The study was approved by the Muhimbili University of Health and Allied Sciences (MUHAS) Institutional Review Board (IRB). A cross-sectional study was then conducted in Kongwa district, one of the six districts in Dodoma region (province), located in central Tanzania. The district is divided into 22 administrative wards, each of which is further sub-divided into 3-5 villages. The population of the district is estimated at 300,000 people. The district is served by one public hospital, three health centres and eight dispensaries.
Sample size calculation considered the national stunting prevalence rate of estimated at 42 % , power 90 % and 0.05 precision yielding 688 as the a sample size. We randomly selected two wards out of the 22 wards in the district. The first one, Mlali, had a population of 19,623 and the second, Ugogoni, had 17,048 people. A sampling frame consisted of households with an under-five child. Once compiled, the sampling frame was subjected to random selection of the required number of households from each village (with the sample size corrected for the sampling design). Equal numbers of households were selected from each ward. Within a household, both the mother and the head of the household were interviewed and anthropometric measurements for under-five children were taken.
Data collection tools
Field data were collected using a pretested structured interview guide that was administered by trained research assistants. The interview guide collected information on the socio-demographic characteristics, ownership of selected assets, distance to the health facility and time to the nearest water source.
Permission was obtained from the mother prior to weighing the child using Salter scales. All the children were weighed wearing minimal clothing and without shoes. The weight was recorded to the nearest 100 grams and the height to the nearest 1 mm . Each day and before any weighing, the scale was adjusted to zero for accuracy and consistency by using a 2 kg weight. A pair of trained research assistants assisted by the mother or guardian used a length measurement board with a sliding headpiece to measure the height of each child. For children under the age of two years as well as children who were unable to stand upright, their recumbent height was determined using a length board placed on a flat and firm surface. Such children were then placed lying supine along the middle of the board. Appropriate records were entered accordingly into the child’s checklist, which was checked for completeness and consistency at the end of the day.
Data management and analysis
The data were checked for completeness and consistency before being entered into the computer using EPI-INFO 3.2.3 database. Data cleaning was done and then followed by the generation of frequencies and descriptive statistics using STATA version 12 . Independent variables included individual and community level variables. Individual variables included the age and education of the mother and the head of household, the number of children in the household and the marital status of the mother. Community (place) variables included distance to the nearest health facility and time to the nearest water source. The WHO Emergency Assessment Tool was used to calculate Z-scores using the anthropometric data of each child. Those with less than minus two (-2) Z-scores were classified as stunted and were coded as 1 and the rest were coded as 0.
From the outset, comparison of percentages of stunted children by categories of the independent variables was done using the chi-square test to determine the significance of the difference observed. A multivariate logistic model that entered all 12 independent variables in one single step was adopted. The multivariate logistic regression determined the Adjusted Odds Ratio (AOR) for the association of the dependent and independent variables controlling for the rest of the variables. Significance level was set at p-value 0.05 or less.
Prior to their participation, the aim and benefits of the study were explained clearly to the parents in addition to their right to voluntary participation. Respondents were also given an opportunity to ask questions for clarification purposes and when it was clear they had understood the nature of the study and their rights, each participant was requested to provide a written consent; only those who consented were recruited for participation in the study.