Study site
The study was conducted in five villages from the semi-arid Dodoma region in Tanzania (Figure 1). Food production in Dodoma is predominantly rain fed. Dodoma region receives rainfall in one season with an average of 350-500 mm rainfall per annum. Dodoma is characterized by a prevalence of highly food-insecure areas. Crops produced include cereals (sorghum, pearl millet and maize), roots/tubers (cassava and sweet potato), legumes (cowpea, pigeon pea, bambara nut, groundnut, chickpea, green gram and lablab bean), oil crops (sunflower, sesame, groundnuts) and fruits (pawpaw, guava, mango, grape, lemon and dates). There is also widespread collection of edible wild fruits and vegetables. The food system in Dodoma is mainly based on cereals with pearl millet as the preferred staple. Groundnuts are normally mixed in most relishes that are used with the main dish. Edible wild products, particularly vegetables and fruits, are important in local food menus [7]. The Chamwino district imports food crops from other regions during deficit months. These foods include maize, beans and pigeon pea. During deficit months imported food such as maize and pearl millet is sold at a price more than three times its price during the months of plenty. This is because there are no structured local markets in the study villages, only small grain and pulse traders. The two regions together account for 70–80% of the types of farming system found in Tanzania [8].
Sampling and sample size
Multistage stratified sampling procedure was used to select respondents. First, purposive sampling was used to select Dodoma region as one of the regions with highest prevalence of malnutrition in Tanzania (36.9%) while national prevalence is 32% [9]. Due to homogeneity in the number of households between villages, five villages were selected, namely: Mzula, Ilolo, Ndebwe, Mvumi-Makulu and Chalula. Second and third stages were stratified based on the information obtained from village record offices. The third stage involved selection of households that met the study criteria.
The total sample size of 660 households was computed using Fisher’s formula, with the prevalence of anaemia in rural areas used as a basis for the determination of sample size [10]. In the fourth stage, due to less variation in the number of population sizes between the villages, the obtained sample size was equally divided into five villages, which gave an average of 132 households for each village where simple random sampling method was applied to select the required equal number of households in all the villages. The inclusion criteria to participate in the study were; a rural household with a mother or a caregiver and a child aged 23 to 59 months. Households excluded in the study were those which did not have a mother or caregiver or a child of that age. All the eligible households were listed from the village registry and subjected to Emergency Nutrition Assessment (ENA) for SMART software for randomization. This led to the selection of the 660 households from five villages that participated in the study.
Protocol
Permission to conduct the study was obtained from the Sokoine University of Agriculture and from Chamwino District Commissioner’s Office. Household heads and spouses were informed of the purpose, objectives and activities of the study by reading to them the information sheet. The participants were required to sign the consent form or apply a thumb print (in ink), marking their consent to participate in the study.
Social, economic and demographic indicators
In the selected households, an interviewer-administered questionnaire was used to collect demographic and socioeconomic data as well as to assess the knowledge of mothers or caregivers in nutrition and practices. Five nutritionists were taught to conduct anthropometric measurements and interview caregivers in order to assess living conditions and nutrition practices.
In addition, we inquired if the child had suffered any major health issues since birth, and if the mothers were concerned about the child's growth and development. To get a better idea of their economic situation, we noted the type of house they lived in (with a focus on the building materials used), if they owned the house, and how much time they spent fetching water. The use of 'assets' or a 'wealth index' as estimates of expenditure and income has been proposed, particularly in developing nations where it is hard to estimate levels of income.
Anthropometric measurements
Height and weight were measured and nutritional status of children was determined. The WHO [11] guidelines and standards were used to define stunting, underweight and wasting. All standard methods of taking anthropometric measurements were carried out in accordance with relevant guidelines and regulations accordingly [12, 13].
The children's ages were collected from their parents and verified using their clinic cards, if accessible. The height (in cm) and weight (in kg) of children were measured. A SECA electronic bathroom scale was used to measure weight to the closest 0.01 kg (A SECA, Vogel and Haike, Hamburg, Germany). A stadiometer was used to measure height (Shorr Productions, Perspective Enterprises, and Portage, Missouri, USA). The child was measured while standing without shoes on a horizontal flat plate attached to the base of the stadiometer, with their heels together, stretched upwards to their utmost length.
Data analysis and statistical methods
The data was analyzed using the Statistical Product and Service Solution (SPSS) software version 20 (SPSS Inc., Chicago, IL, USA). The ENA for SMART 2011 (www.nutrisurvey.de/ena2011/) was used to categorize the study children into nutritional status categories by converting anthropometric measurements into z-scores, such as weight for age Z scores (WAZ), height for age Z scores (HAZ), and weight for height Z scores (WHZ). Stunting, underweight, and wasting in children were defined using these criteria, which were compared to WHO norms and standards (WHO 2006).
While controlling for other variables, the net effects of each independent variable were evaluated using logistic regression multivariate analysis. In respect to independent factors in the models, the odds ratio was utilized to quantify the risk (increased or decreased) of stunting. When the P value was less than 0.05, significance was evaluated. A logistic model was run. Child stunting was the dependent variable in this model. Stunting (a child's height in relation to his or her age) is a common marker of malnutrition and is seen as a good indicator of poverty, showing insufficient food consumption over time [14]. Stunting is a typical symptom of chronic malnutrition that is linked to environmental and socioeconomic factors [15]. Household size, cultivated land size, gender of the child, age of the child, literacy status of the mother, use of iodized salt, body mass index of the mother, breastfeeding duration, distance to a water source, region of residence, marital status of the mother, age of the mother, and gender of the mother were the independent variables considered in the regression.
Ethical considerations
Permission to conduct the study was obtained from the Chamwino District Commissioners’ Office. Household heads and spouses were informed of the purpose, objectives and activities of the study. The household representatives were required to sign the consent form, marking their consent to participate in the study. For participants who couldn’t read the informed consent form, the document was orally presented to them and they were required to apply a thumb print (in ink) to give their informed consent. All participants of this study gave the informed consent. Ethical clearance was obtained from the Sokoine University of Agriculture Ethics Committee. The SUA Senate Ethical, Research and Publication Committee approved the study. All methods used in this study to measure weight and height and other antrhopometrics were carried out in accordance with relevant guidelines and regulations.