Study area and setting
Community based cross-sectional survey was carried out among 6–23 months old children residing in 6 villages of Damot Sore District, situated in Southern Nation Nationalities and Peoples Region (SNNPR), which is 326 km from the capital city of Ethiopia. All children in the aforementioned age group who stayed in the study area during the time of the interview were qualified to take part in the study.
Sample size and sampling procedure
A formula for estimation of single population proportion was used to calculate the sample size. The following criterions were applied to estimation of the sample size; 95% confidence level, 5% error of margin, and 66.6% prevalence of anemia among children aged 6–23 months old – taken from a previous study in a rural area in northern Ethiopia [8]. The estimated sample size was then adjusted for a non-response rate of 5% and multiplied by the design effect of 1.5 to obtain a final estimate of 498, the sample size used for this study.
Study participants were selected by a multi – stage sampling technique. From a total of 20 rural villages, 6 were randomly selected using a lottery method.
Sampled number of children per village were proportionally allocated based on their number of households. Finally, using family folder which is found in community health information system (CHIS) of village health posts as a sampling frame, households with children 6–23 months old were selected by simple random sampling method. For a household that have twins or more than one child resident there, one of them was selected by using lottery method. In case of absence of an eligible child in a given household, a substitution was made by a child in the next household. Children with physical deformities of limbs, spine, suffering from diseases and have mental defects were excluded. In addition to this, children who had received blood transfusion and anti helmenthiasis prior (two months) to data collection were excluded.
Data collection
Data were collected using the questionnaire adopted from previous studies [4, 9] and pre-tested before the start of this study. The questionnaire was first prepared in English language and then translated to Amharic language. Prior to collection of data, the purpose of this study was explained to the study units; their consent to participate was sought and was also informed that their participation in the study was totally voluntary.
The response from the mother/caregiver was recorded after the data collectors read out the questions loud. Date of interview and date of birth was used to calculate age of the child, because the year of birth is inaccurately announced oftentimes. The nutritional status for all children aged 6–23 months old was assessed by taking Anthropometric measurements.
Data collectors and measurements
Anthropometric data for this study was collected by six skilled and trained data collectors who administered the questionnaires. Two supervisors closely supervised the process of data collection. Nutritional status was assessed by taking anthropometric body measurements of the children. Length of a child was measured in a recumbent position to the nearest 0.1 cm by using a portable board provided by UNICEF (United Nations Children’s Fund) with an upright movable wooden base. Anthropometric measurements were converted to z-scores of indices using WHO Anthro software [10].
Laboratory investigations
Hemoglobin count and malaria status of children were investigated. Hemoglobin was measured from capillary blood by aseptically collecting blood sample from the middle finger of study participants, then the analysis have been done by using Automated HemoCue analyser (HEMOCUE Hb 301, HEMOCUE AB, ANGELHOLM SWEDEN) machines and the results were immediately recorded in the field in terms of g/dl. After adjusting the hemoglobin concentration for changes in the altitude and smoking individual within a household, the results were categorized based on the WHO cut off point, which categorizes a child as anemic if the hemoglobin count is less than 11.0 g/dl [11]. Malaria test was done using rapid diagnostic test (RDT) kit, which was commonly used to assess the status of malaria in the community [12]. Blood test for malaria was collected by finger puncture and the result was recorded as positive or negative with regards to species specification.
Data quality control
Three day training was given for data collectors about study objective, interview techniques, anthropometric measurements and ethical issues during data collection. Rapid diagnostic malaria test results were compared with blood film result by microscope. Standard operating procedures and manufacturer’s instructions were strictly followed starting from sample collection up to result reporting for laboratory activities.
The questionnaire was pre-tested on similar setting outside the study area before the collection of actual data. The principal investigator carefully monitored the data collection process.
Quality of the measurements were ensured by maintaining consistency of anthropometric measurement, data collectors were tested using ENA for SMART software before starting data collection.
Standardization: all children were measured without any shoes and clothes were taken off.
Multicollinearity for independent predictors of stunting and anemia were checked and Crombach’s alpha was checked for household wealth. Data cleaning were done and outliers were identified and managed properly before the analysis.
Data management
The data management were done by using three statistical softwares. During the data collection, completeness and uniformity of the data were checked daily before entry.
The data were first entered into EpiData V.3.1 statistical software for coding. Afterwards the data were transported into the software WHO Anthro, where length-for-age Z-scores were computed and further checks done to ensure that flags resulting from wrongly entered data were corrected. After the initial cleaning, all the z-score values which remained as irregular were cleaned from the file and excluded from further analyses. The cleaned file was then exported to SPSS version 21.0 for further analyses.
Statistical analyses
Bivariate and multivariable logistic regression was used to examine the association between stunting, anemia and the explanatory variables. From the binary regression models, independent variables which were associated with the outcome at p-value less than 0.25 were selected as candidate for inclusion in the multivariable logistic regression models. Statistical significance was set at p < 0.05 and 95% confidence interval.
Operational definition
Stunting: is defined as length-for-age Z-scores below minus two <− 2 Z score or Standard deviation of the reference population of World Health Organization (WHO) Multicentre Growth Study. Severe stunting is defined as LAZ scores below minus three <− 3 Z score or Standard deviation of the reference population of WHO Multicentre Growth Study [10].
Anemia: A child is considered to be anemic if the hemoglobin count is less than 11.0 g/dl against the WHO reference range [11].
Poor DDS: dietary diversity of less than 4 food categories.
Good DDS: dietary diversity of more than or equal to 4 food categories.
Poor breast feeding practice: failed to breast for at least 8 times per day or inappropriate baby position or switching to the next breast without finishing.
Good breast feeding practice: breast feed for more than or equal to 8 times a day or appropriate baby position or switching to the next breast after finishing one.