The current study was conducted in Shinile Woreda, ESRS, between February and March, 2014. The ESRS is one of the nine regional states that constitute the federal democratic republic of Ethiopia . Geographically, the region occupies a large area and falls in the eastern and southeastern part of the country with land mass area of about 350,000 km2 and an estimated total population of 4,445,219 people . 86.1 % of the population resides in rural areas while only 13.9 % reside in urban areas . Shinille Woreda has a latitude and longitude of 09°41′N and 41°51′E with an elevation of 1079 m above sea level. Based on figures published by the Central Statistical Agency, the Woreda has an estimated total population of 113,630, of which 46.6 % are men and 53.4 % are women; 23.91 % of its population is urban dwellers, and the Woreda is composed of 30 kebeles. In this Woreda, 80 % of the inhabitants are pure pastoralists, while 15 % agro pastoralists and 5 % of the community engaged in petty trading and other activities in urban town. The Woreda is characterized by arid and semi arid climate with an annual rain fall ranges between 450 to 550 mm/year. The annual temperature is varying from 30 to 37° centigrade.
Study design, source population and study participants
A community-based cross-sectional study design was used. The source population was all children under five years of age residing in three kebeles, which were selected by simple random sampling. Study participants were all children 6–59 months of age who were selected by simple random sampling from eligible children in the selected households. Children were not included if they were guests, if they were seriously ill for sickness and if their mothers did not consent to participate in the survey.
Sample size and sampling technique
The required sample size was calculated using the formula required for determination of sample size for estimating single proportion. Based on the prevalence of malnutrition among under five children (29 %) , and with additional assumption of 95 % confidence interval, 5 % margin of error, a design effect of 2 due to multi-stage sampling and 10 % non-respondent rate in our estimate, a total sample of 697 were needed. During sampling, 3 out of the 30 kebeles in the woreda were selected using simple random sampling technique and the calculated sample was proportionally distributed to the selected kebeles based on their number of households. Then, from each of the selected kebeles, households were selected using systematic random sampling technique. Finally, from all the eligible children in a household, only one was selected by simple random sampling for the anthropometric measurement. In addition, the mothers of all selected children were interviewed. But in the absence of eligible child in a given household, a substitution was made by a child in the next household.
Data collection and quality control
Data were collected using structured questionnaire and anthropometric measurement (Additional file 1). Twelve data collectors and two supervisors were recruited from health centers and health posts in the woreda. Training was provided for data collectors and supervisor for two days.
The questionnaire was used to collect quantitative data on variables pertaining to the socioeconomic and demographic characteristics of the participants. It was first designed in English based on information from other literatures developed for similar purpose [25, 31]. After adopting to the local context, the questionnaire was translated to Somali (the local language of the study area) (Additional file 2). Moreover, the questionnaire was pre-tested on randomly selected individuals from the survey area and these individuals were not participated in the main study. During the pre-test, the questionnaire was assessed for its clarity/understandability, reliability, sensitivity of the subject matter and for cultural acceptability in the area.
The mothers/care takers of the children were interviewed to provide answers to questions other than child anthropometry. Each face to face interview was made by a house-to-house visit and the participants were interviewed in their local language. All the interviews, measurements and testing were conducted at the residences of the study participants. On daily basis, collected information was reviewed and possible errors were returned to the collectors for correction.
To assess the physical growth and nutritional status of the children, measurements of height and weight were taken of all of the children and these measurements were taken during the home visit. These anthropometric data were collected using the procedure stipulated by the WHO  for taking anthropometric measurements. Before taking anthropometric data for the children, their age was determined in order to ensure the target population. Local events were used to establish the birth period. The mothers were asked whether the child was born before or after certain major events until a fairly accurate age is pinpointed. If age couldn’t be determined accurately, a height of 65–110 cm was considered as proxy indicators.
Body length of children aged up to 23 months were measured without shoes and the height was read to the nearest 0.1 cm by using a horizontal wooden length board with the infant in recumbent position. However, height of children aged 24 months and above was measured using a vertical wooden height board by placing the child on the measuring board, and child standing upright in the middle of board. The child’s head, shoulders, buttocks, knees and heels touching the board.
Weight was measured by electronic digital weight scale (salter model 235-6 s) with lightly clothing and no shoes. Calibration was done before weighing every child by setting it to zero. In case of children aged below two years, the scale was allowed weighing of very young children through an automatic mother-child adjustment that were eliminated the mother’s weight while she standing on the scale with her baby.
Edema was checked and noted on data sheet because children with edema were severely malnourished. In order to determine the presence of edema, normal thumb pressure was applied to the two feet for three seconds whether a shallow print or pint remains on both feet when the thumb is lifted.
To identify retrospective morbidity of children, mothers were asked about any occurrence of illness during the past two weeks. Enumerators probe to confirm nature of illness based on operational case definition and was asked to identify occurrence of measles in the past one year. In addition, vaccination status of children was checked by observing immunization card and if not available mothers were asked to recall it. BCG vaccination was checked by observing scar on right (also left) arm.
Variables of the study
The main outcome variables were: prevalence of stunting, wasting and underweight as determined by anthropometric measurements of weight and height taking age and sex into consideration. Five categories of independent variables were included in the data analysis. 1) Socio-economic and demographic variables; such as head of the family (father or mother), family size, income, ethnicity, religion, parental literacy (able to read or write) and occupation. 2) Child characteristics; Age, Sex, birth order, place of delivery, types of birth, birth size/weight, breastfeeding status, gestational age and morbidly status (fever, measles, diarrhea). 3) Child caring practices; feeding, hygiene, health care seeking and immunization. 4) Maternal characteristics; age, number of children ever born, antenatal care (ANC) visits, health status during pregnancy, use of extra food during pregnancy/lactation and autonomy in decision-making on use of money. 5) Environmental health condition; Water supply, sanitation and housing conditions.
Data processing and analysis
First, the data were checked manually for completeness and consistency. Then, it was coded and entered in the computer using EPI-INFO3.5.1 software and then sex, age, height and weight transferred to ENA for SMART 2007 software to convert nutritional data into Z-scores of the indices; Height-for-Age Z-scores (HAZ), Weight-for-Height Z-scores (WHZ) and Weight-for-Age Z-scores (WAZ) using the National Center for Health Statistics reference population standard of WHO. A child was considered stunted, wasted or underweight if the corresponding HAZ, WHZ and WAZ -scores were less than −2. When the measures of HAZ, WHZ and WAZ were less than −3, the child was considered severely stunted, severely wasted and severely underweight, respectively. Then, the data were exported to SPSS V.16 for analysis. Descriptive summary using frequencies, proportions, graphs and cross tabs were used to present study results. Bivariate analysis was computed to determine the association of malnutrition and associated factors. Statistical association was checked by 95 % confidence interval (CI) and crude odd ratio (OR) and the significant variables (p-value < 0.25) observed in bivariate analysis were subsequently included in multivariate analysis. Finally, results were reported as statistically significant whenever p-values were less than 0.05 at 95 % CI. OR was used to report strength of association between background variables and the target outcome variables.