Study site
We conducted this cross sectional study among children aged 6 -59 months in Isingiro district, located in the Ankole sub-region, south western Uganda. The district has a population of 492,117 people in 102,967 households. Approximately, 23.2 % of the population is children aged 0-5 years. The district has a high fertility rate at an average of 6.2 children per mother (12). About 80 % of the population derive their livelihoods from subsistence farming. Key among the crops grown in the district are beans, and iron bio fortified beans were introduced in 2016 (10), as a way of addressing the high prevalence of anaemia which stands at 39.8% (13). The district experiences equatorial type of climate coupled with long dry seasons, which sometimes causes food insecurity among households.
Sample size and sampling procedure
We calculated our sample size using Kish Leslie (1964 ) formula for cross-sectional studies since our expected outcome is a categorical variable (anaemic and non-anaemic) reported as a proportion (14). We used anaemia prevalence of 31% from a previous study (15) and a 5% margin of error and adjusted for non-response at 5% (16) and design effect of 1.5 (17), giving us the final sample size of 519 participants.
We selected one rural and one urban sub counties using simple random generated numbers with ENA SMART (emergency nutrition assessment) software. Selection of a rural and an urban area ensured that there was a fair representation of the study area.
Secondly, we randomly selected two parishes from each sub county. The parishes in each sub county were numbered consecutively and the random selection done using ENA SMART.
A total of 519 participants were eligible for this study. We sampled 503 participants but two households were excluded because they were headed by minors. We therefore collected data from 499 children. However, two participants were excluded from the analysis because blood samples were not drawn for haemoglobin determination because they had fever.
Variable measurements
The outcome variable was absence or presence of iron deficiency anaemia; with the latter defined as haemoglobin levels (Hb) <11g/dl.
Independent variables included social demographic characteristics of the mother or primary caretaker of the child such as highest-level of education attained which was categorised as none, primary, secondary and tertiary; main occupation which was categorised as subsistence farming, salaried jobs or business; and age (list categories). The child’s data collected included age, sex (male/female) and disease history in last two weeks. Data on disease history was collected because we wanted to determine health status of the children and rule out malaria and infections that are usually characterized by high fever, diarrhoea, which could alter the child’s heamoglobin status. Intake of iron bio fortified beans: which was defined as having consumed a meal containing iron-bio fortified beans in the 24 hours preceding the study. It was categorized as consumed or did not consume iron bio fortified beans. Intake of iron rich foods using dietary diversity: was defined as having consumed food from four or more foods in the 24 hours preceding the study. A child was considered to meet minimum dietary diversity if he/she consumed food from four or more food groups (19).
Data collection
Social demographic
Electronic questionnaires designed in ODK were pre-tested and used to collect socio-demographic data in a face-to-face interview. Data collected from the mother/caregiver were highest level of education attained, occupation, age, household size, marital status, and relationship with the child. The child data collected were health status, sex, and age.
Blood samples
Blood samples were collected by trained and registered medical laboratory personnel. The finger was cleaned with alcohol pads to sterilize the area before pricking using Blood lancets penlet and a drop of blood from the pricked area was drawn into a micro cuvette and fitted in the HemoCue. Haemoglobin analysis was carried out on-site with quality controlled battery operated portable HemoCue analyser 301 and results were recorded to the nearest 0.1g/dl. Haemoglobin levels were categorized according to WHO classification of anaemia among children as severe <7.0 g/dL, moderate 7.0-9.9g/dl, mild 10.0-10.9g/dl and no anaemia ≥11g/dl (1). The results were communicated to child’s mother or caretaker.
Consumption of iron bio fortified beans
Intake of iron bio fortified beans was determined by using 24-hour food frequency questionnaire and individual dietary diversity score. Respondents were asked to describe all foods (meals and snacks) that a child ate or drunk during breakfast, lunch and supper in the previous 24 hours(20). All the foods, snacks and drinks mentioned were written down. When composite dishes were mentioned, the respondent was asked to list all the ingredients. After the respondents had finished recalling, research assistants probed to find out if there were any other meals not mentioned(19). In addition, if beans had been mentioned the respondent was asked which type of beans was consumed. Iron bio fortified bean charts and bean seeds were shown to respondents to confirm the type of beans eaten.
Dietary diversity
A food dietary diversity questionnaire consisting of 60 foods from 15 different food groups was used to collect data. To determine the dietary diversity score, foods were categorised into the following seven groups: i. grains, roots and tubers ii. legumes and nuts iii. dairy products (milk, yoghurt, cheese) iv. flesh foods (meat, fish, poultry and liver/organ meats) v. eggs vi. vitamin-A rich fruits and vegetables vii. other fruits and vegetables.
A child who ate from at least four of the above 7-food groups meets a minimum dietary diversity score (19, 21). The cut-off was selected because it was shown to be associated with better quality diets for both breastfed and non-breastfed children. The food groups were carefully selected using infant and young children feeding guide to include iron rich foods from both plant and animal sources (21).
This method was good because the period was short and the participants would easily recall. It was more appropriate for measuring dietary intake and diversity in individuals at the community level. It also considered all foods eaten including mixed dishes.
Analysis
Modified Poisson regression was used to measure association between iron deficiency anaemia and consumption of iron bio-fortified beans and other independent variables. The variables with P valve of ≤ 0.2 were selected for multivariate analysis to include variables that would be of statistical significance or could have an influence on the outcome of interest. Modified Poisson regression was used because prevalence of anaemia was higher than 10% (26.3%).
Multivariate analysis was conducted using modified Poisson regression model with robust error variance to estimate prevalence ratios (PR) as a measure of association between anaemia and intake of iron bio fortified beans and other independent variables. A forward stepwise model was used. The independent associations were determined at 95% confidence interval and variables with P-value less than 0.05 were considered significant.