The study was conducted in Pajule sub-county (20 56’ 23” N and 320 56’ 38” E) located in Aruu North constituency, Pader District in northern Uganda [Fig. 1]. Pajule sub-county consists of six parishes and has a population of 22,713, with 4,050 of these being children below 5 years [11]. Like in other parts of northern Uganda, poverty level is higher than the national average, due to a combination of factors like the prolonged civil war that affected the entire northern region, cattle rustling by the Karimojong, and marginalisation that dates back to the colonial era [9]. The majority of the households derive their livelihood from subsistence farming; only 27% depend on earned income [13]. Water coverage has reduced from 57% when the population was in camps to only 38% as the communities returned to their homes [14].
[Figure 1]
Study design and data collection
A cross-sectional survey was conducted in April 2018 in four randomly selected households from four parishes out of the six parishes in Pajule sub-county. According to the 2014 National Housing and Population Census, Pajule subcounty has approximately 4000 households, and so a sample size of 351 households was estimated using the Krejcie and Morgan table [15]. However, due to logistical constraints, non-response or unavailability of targeted respondents at the time of survey, we only sampled a total of 244. The number of households in each parish was determined using probability proportional to size, and from each parish, at least two villages were randomly selected using a random number. The list of households in each village was obtained from the respective local council chairpersons. Only individuals from households where the mother or caregiver was present and had a child < 5 years old were interviewed. In cases, where there were more than one child < 5 years in the same household, index child was selected by lottery method. Trained research assistants administered semi-structured questionnaires based on the World Health Organization (WHO) guidelines [16]. The dependent or outcome variable was the presence of diarrhoea among children aged < 5 years within the 14 days before the survey. This was evaluated by asking the mother or caregiver if the child involved in the study had suffered from diarrhoea within 14 days before the study. Independent variables included socio-demographic, socio-economic, environmental and behavioural factors. Socio-demographic and economic characteristics included age of the child, number of children under the age of five in the household, family size, age of the child’s caregiver, sex of the child, the income status of the family, and the mother’s or caregiver’s education level. Environmental factors included type of water source, availability of animals’ in the homestead, presence of animals’ houses, child’s stool disposal practice, availability of latrines, ownership of latrine, hand washing practices, availability of kitchen, household’s environmental cleanliness, and presence of utensils’ drying racks. The behavioral characteristics included source of drinking water, boiling of water before consumption, frequency of warming cold food, weaning age, and age of food supplementation.
Data analyses
Descriptive analyses using frequency and percentages were used to summarize the independent and dependent variables. To obtain the associations between diarrhoea among children and risk factors such as socio-demographic and socio-economic profile, environmental and behavioral factors, we use multivariable logistic regression. The adjusted odd ratios [AORs] of having diarrhoea with 95% confidence interval [CIs] and P value <0.05 were used to describe associations. First, we conducted univariate analysis to determine the associations between diarrhoea and other associated factors using chi-square and binary logistic regression. Eight variables with p-values less than 0.05 in bivariate analysis were included in the final multivariable logistic regression. All analyses were done using IBM SPSS for windows version 25.