This cross-sectional study was performed using the baseline data for a large effectiveness study on the use of multiple micronutrient powders (MNPs) within a local CF program on iron status, morbidity, and children’s growth. Methods and findings of the MNP effectiveness study are described in detail elsewhere [12].
Study area and population
The baseline data collection took place in the predetermined nine districts from Oromiya and Southern Nations, Nationalities, and Peoples (SNNP) regions of Ethiopia between March and April 2015. Both regions were selected because of their similar characteristics in food security, child health, and nutritional status; explicitly stunting (37% and 39%) and wasting (10.6% and 6.0%) in Oromiya and SNNP regions respectively [2], and infant feeding practices [13].
The study population consisted of young children 6–11 months of age [12]. The study intended to assess whether there will be a difference in IYCF practices between boys and girls, particularly in the introduction of solid, semi-solid, or soft foods which usually need to be initiated at 6–8 months [12]. This difference will lead to a difference in nutritional status. In addition, the objective of the intervention study was to assess the effectiveness as well as risks and benefits of a low-dose (6 mg iron per serving every other day) iron Micronutrient Powders (MNP) on iron status, morbidity, and growth of Ethiopian infants and young children [12]. We chose this age group to ensure that all children continued to be eligible to receive the program intervention for the entire duration of the follow-up. Details on the study site and sample selection are described elsewhere [12].
Operational definition
Stunting is defined as low height-for-age and it results from chronic undernutrition.
Wasting is defined as low weight-for-height. It often indicates recent and severe weight loss, although it can also persist for a long time.
Sample selection
We used a three-stage sampling for sample selection. At the first stage, intervention and non-intervention districts were purposively selected which were 3 intervention districts and 3 non-intervention districts which make 6 districts per region in total 12. Out of these 12 districts, three districts were excluded based on the status of the grain-bank program leaving nine districts for the study. Once the districts were chosen, the sub-district (second stage) within the district were selected using inclusion criteria [12]. A total of 35 sub-districts were identified. At the third stage, the selection of households within the sampled sub-districts was performed using criteria including a child 6–11 months, a child not having any serious illness that changes his/her food intake, and a mother willing to stay in the sub-district during the study period. Before screening for admission, all children below 1-year age in the selected sub-districts were listed by Health Extension Workers. Using the criteria, 65 eligible children (6–11 months) per sub-district; a total of 2356 children were included.
A total of 2356 children from 35 sub-districts/clusters (the smallest administrative unit in Ethiopia) of 9 districts were screened and admitted to the study [12]. A child could participate in the study if he/she was ≥ 6 and < 12 months old on the recruitment day and living in one of the selected sub-districts. Participating children also had to be free from chronic conditions such as metabolic or neurological disorders that might impact their health (e.g. mental retardation). Exclusion criteria included the presence of serious disabilities that would affect normal growth and development. In addition, children with a severe or protracted illness for which continuous medication is required and children with severe undernutrition (weight-length-z score (WLZ) < -3 SD) were excluded from the study and referred to the nearest health facility. In total, 320 children were excluded, and the analyses were performed on data for 2036 children.
Data collection and measurements
Procedures
Three data collectors were assigned to work on a study area of three sub-districts as a field team. Thirty-six data collectors and six field supervisors were trained in the administration of questionnaires (IYCF, 24 h. recall, and morbidity) and anthropometric measurements, including standardization of data collectors (see supplemental file). After nine days of training on methodological procedures and quality assurance, the questionnaires were tested in a pilot group and adapted based on the received feedback from the survey team. The questionnaires were translated into local languages (Oromifa and Amharic) and back-translated to English to ensure the quality of the translation. The data collectors’ measurements were standardized to ensure that the inter-observer variability was within tolerable limits. Supervisors received additional training on teamwork and monitoring and supervising the data collection process. The field teams were provided with training and data collection manuals, a chart for calculating age in months, an event calendar, and a World Health Organization (WHO) classification table for Weight for Height (WHZ) to identify severely malnourished children. Data were collected at the sub-districts’ health posts by interviewing the child’s mother/primary caregiver and taking anthropometric measurements. The supervisors monitored the measurements to ensure the quality of the data.
Anthropometric measurements (length and weight) were taken following standard procedure [14]. Weight was measured using the UNICEF Seca 874 U electronic scales (UNICEF Supply Division, Copenhagen, Denmark) with 100 g precision calibrated daily with a known weight, and height was measured on UNICEF’s standard measuring board (precision of 0.1 cm). All children were measured lying down. Measurements were taken in duplicate and repeated a third time if the difference between the first two was more than 0.5 cm or 0.5 kg. IYCF practices and morbidity status were assessed using a questionnaire based on WHO recommendations to collect data for the IYCF indicators [15]. A 1-day non-quantified 24 h dietary recall was collected, including information on the source of food, method of preparation, and meal description to assign all ingredients to the respective food groups consumed by the child in the previous 24 h period.
Data processing and analysis
Data processing
All the questionnaires were manually checked for completeness before data entry. Data were coded in duplicate and analyzed using SPSS (Version 22.0 for Windows, IBM, New York, USA). The data were cleaned for inconsistencies and missing values. If inconsistencies and missing values could not be resolved by checking the original questionnaires, those data were excluded from further analysis. Children’s age was entered as the date of birth provided by the caregivers. Length-for-age z scores (LAZ) and WLZ z scores were determined using the WHO Anthro software version 3.2.2 [16] based on the WHO reference population (2006). Stunting was defined as LAZ < -2 of the standard deviation (SD) and wasting was WLZ z scores < -2 SD. IYCF indicators were calculated following the UNICEF guidelines [17]. For this study, we defined as ‘timely introduced to CF’ a) children aged 6–8 months who were fed breast milk and had had at least one solid or semi-solid food the previous day or b) children aged 9–11 months who had a recall age of the first introduction of CF between 6 and 8 months of age. The seven food groups described by the WHO [17] were used to classify foods consumed, namely: 1) grains, roots, and tubers; 2) legumes and nuts; 3) dairy products; 4) flesh foods; 5) eggs; 6) vitamin A-rich fruits and vegetables; 7) other fruits and vegetables. Minimum dietary diversity was defined as the consumption of four or more food groups from the seven food groups [17]. Minimum meal frequency was defined as the consumption of 2 or more (at age 6–8 months), 3 or more (at age 9–23 months) solid or semi-solid feeds for breastfeeding children, or 4 or more solid or semi-solid or milk feeds for non-breastfeeding children at age 6–23 months [17]. A minimum acceptable diet was defined as a combination of minimum dietary diversity and meal frequency [17]. Basic drinking water and adequate sanitation facilities were defined according to UNICEF and WHO’s joint monitoring programme water, sanitation, and hygiene (WASH) targets and indicators post-2015 [18]. Full anthropometric data (age or length or weight) were missing for about 8% of children and the WLZ z score of some of the children was below entry point (< = -3.0). Thus, children with incomplete data on nutritional status and the flagged cases were therefore excluded from our analyses.
Statistical analysis
All continuous data were checked for normality by investigating the skewness and kurtosis of the distribution. For the linear regression, we used a graph of standardized residuals against each of the predictor variables in the regression models. We also used graphs to test the homogeneity of variance of the residuals by plotting the residuals versus predicted values.
Child characteristics, i.e. age, height, weight, LAZ z score, WLZ z score, stunting, wasting, and IYCF indicators, mother/caregiver characteristics including mother’s age, marital status, maternal education and occupation, and household characteristics, i.e. WASH indicators and region of residence, were compared between boys and girls using Chi-square tests for categorical variables and Student t-tests for continuous variables.
Binary logistic regression was used to investigate the association between potential determinants vs stunting and wasting. Variables associated with the outcome at p < 0.2 were selected for multiple logistic regression, stratified by gender, and adjusted for age, sub-district/ cluster, and mother’s characteristics. The interactions between each variable and gender were tested with cross-product terms. For continuous variables (LAZ z score and WLZ z score), multiple linear regression was performed including independent variables associated with the outcome in unadjusted analyses at p < 0.2. A p-value of < 0.05 was considered significant. For the tests of interaction terms, a p-value of < 0.2 was considered relevant [19].