Food Consumption and Iron Status Survey in Two Provinces of Rural Burkina Faso

Objectives of the study

The main objectives of the study were:

• To determine iron, zinc, polyphenolic compounds, and phytate content in raw sorghum grains and cooked sorghum based dishes;
• To estimate the prevalence and level of sorghum consumption in women and preschool children;
• To determine the percent contribution of Sorghum to total vitamin A, iron and zinc intake;
• To assess the prevalence of inadequate intakes of vitamin A, iron and zinc as defined by the cut-point method using Estimated Average Requirements for women and children according to sex and age;
• To determine the prevalence of micronutrient deficiencies using international (WHO) cut-offs for serum retinol, serum ferritin and serum zinc.

Two secondary methodological objectives were added because of their scientific interest and of their potential practical implications:

• To assess the performance of dietary diversity scores in reflecting adequacy of micronutrient intakes;
• To assess the performance of questionnaires on expenditures at the household level in estimating food consumption of household members.

Setting

The study has been conducted in the "Boucle du Mouhoun" and "Centre-Ouest" regions located in Western Burkina Faso. These regions were selected based on a combination of health, agriculture, living conditions and demographics criteria, which included data on sorghum production, household consumption and prevalence of malnutrition.

The province of Sourou in the "Boucle du Mouhoun" region and the province of Sanguié in the "Centre-Ouest" region were chosen purposely based notably on available information on sorghum production or consumption and on some of the principal investigators general knowledge of the field.

Study design

It was a cross-sectional survey in two rounds: a first round during the lean period (July-August 2010) characterized by very low food availability in Burkina Faso; and a second round during the period of greatest food availability, immediately after harvest (November 2010-January 2011). To enhance the study power, the same individuals were surveyed during these two rounds.

The study had two main components: the "food consumption" component and "lab analysis" component.

The food consumption component included:

• A quantitative 24 hour recall of dietary intake on a random sample of preschool children aged 3-5 y and their mothers
• A qualitative dietary diversity questionnaire for both the children and the mothers, administered on the same day as the 24-h recall but by a different enumerator
• A repetition of the 24 hour recall on non consecutive days using a randomly selected subsample of approximately one third of the original sample
• A questionnaire about income and expenditure on a subsample of approximately one other third of the households (i.e. not the households having a repetition of the 24-h recall)
• Anthropometric data (weight, height and MUAC) collected for all children and their mothers
• A socio demographic questionnaire and morbidity recall for all children and their mothers

The laboratory component included:

• Collection of blood samples from a randomly selected subsample of mothers and children (one third of the original sample) for vitamin A, iron and zinc biochemical indicators as well as haemoglobin and acute phase proteins (C-reactive protein and alpha-1-glycoprotein as markers of inflammation); this blood sampling took place only during the second round of the survey (November 2010 - January 2011).
• Collection of raw sorghum and cooked sorghum dishes samples for determination of iron, zinc, polyphenolic compounds and phytate content; food sample collection took place during the two rounds.

Target population

Women and preschool children were selected because their micronutrient requirements are highest due to their needs for reproduction and growth, respectively. The age range for targeted preschool children was 36-59 months for several reasons: Firstly, we wanted to be sure to exclude breastfed children since the measure of breast milk intakes are very complicated to implement on the field; Secondly, starting at the age of 36 months, most children are used to eating from the common family dish and this made the recall easier, while keeping the number of individual recipes lower. Thirdly, venous puncture was less problematic for children aged 36 months or more.

Sample size for dietary intake: using food consumption data from a survey conducted among women of reproductive age in Ouagadougou, one determined that for a reasonable hypothesis of a coefficient of variation of sorghum intakes of 0.60, with a precision of 0.10, a type I error of 0.05 and supposing a design effect of 1.5, the required sample size was 207 subjects. Finally we decided to set the sample size at 240 households per province and per round, to account for lost of follow-up. The repetition of 24 hour recalls was perfomed on 3 out of 8 subjects.

Sample size for the biochemical indicators: Given a type I error of 0.05, we calculated hypothetical sample sizes for different values of the prevalence of micronutrient deficiencies (30, 40 or 50%) and different desired precisions (either 0.05, 0.075, 0.10, 0.125 or 0.15) and also taking into account different hypotheses about the survey design effect. A huge sample size would have been required to get a precision less than 0.10, depending on the design effect, if data had to be representative at the individual level (women and children) in each province. According to constraints in logistics and financing, we decided to limit the sample size to 90 women and 90 children in each province.

Sample size for sorghum analysis: Five samples of each type of sorghum (red, white, hybrid) and 5 samples of each type of sorghum-based food (paste, gruel) were analysed for each province at each round. This adds up to 30 samples of each class of sorghum and of sorghum-based food in each province at each round, then a total of 120 food samples. However, given the potentially high variability of nutrient density in some food samples, each analysis was performed on a pool of 6 samples of the same type, coming from close villages belonging to a common health area.

Location and sampling procedure

In each province, a multistage sampling procedure was used:

• Firstly a list of all health centres in each province was made and the population size for corresponding health areas was obtained from heath districts and/or administrative data. Five centres were randomly selected in each province, with a probability proportional to size (of population in health areas).
• The same proportional to size sampling technique was used to select 6 villages in each health area, leading to a total of 30 villages.
• Finally, 8 households were selected in each village, from the list of eligible households which was obtained by a census performed by enumerators a few days ahead of the data collection.
• Among the 8 HH included a further random selection was made to determine which of them took part in the repetition of 24-h recall (3 HH) and in blood sampling (3 HH).

Data Collection

Anthropometric data were collected for all study participants (mothers and children) using standard WHO procedures Food intake was assessed by 24-H recall using the multiple pass method. A set of standard recipes was prepared (by observation) and other individual recipes were investigated directly. A local Food Composition table was built by compliling information from available food composition tables (from Mali, FAO, USDA).

The food sampling was done in households selected for this purpose in each village. In these households, about 1 kg of sorghum (white or red) was sampled in clean new polyethylene plastic bags to avoid any contamination. For meals sampling, the selected households were asked to prepare a sorghum based dish (tô or porridge) according to their habits. All samples were put into icebox for transportation to the IRSS laboratory where extraction and analyses was done.

The blood collection was conducted at health centres in cool places with subdued light and conducted by a trained health professional. Care was taken to avoid any contact between the samples and surrounding dust, sweat, or other possible exogenous sources of contamination. The sampling technique was a venous puncture of 10 mL of blood.