A two-stage cluster survey was conducted from the 16th to the 21st of July 2002 in Nyaminyami district. The target population comprised children in the 6-59 month age group. This sub-group is more vulnerable to nutritional stress and is used as a proxy indicator of the nutritional status of the entire population. In line with international recommendations9 on cluster surveys, 30 clusters of 30 children each was studied, which resulted in a sample size of at least 900 children.
A two-stage cluster sampling procedure was applied.
- First level of sampling: selection of the clusters
Clusters were randomly drawn from geographical units by applying the probability proportional to population size (PPS) sampling technique using Cosas Software (version 4.41). This technique ensures that each unit's chance of being selected is primarily dependent on the relative population size but without introducing bias against any of the units. PPS sampling requires the grouping of the population in the smallest available geographical units. In the area of interest these units were villages.
2. Data collected
- Second level of sampling: selection of the children
Once the boundaries of the selected cluster were identified with the help of a community member or survey team member knowledgeable on the area, the survey team went to its approximate centre and picked a random direction. As the homesteads were too scattered it was not possible to count them from the centre to the border of the cluster in the selected direction and select the first homestead to be visited randomly. Instead the survey started at the nearest household from the cluster centre in the selected direction. Then the subsequent households were chosen by proximity. All eligible children found in a household were included in the survey. When possible, temporarily absent children were located and examined. When polygamous households were encountered, one of the wives was selected at random.
The gender of the child selected was recorded.
When birth records or other documents were available, the date of birth was recorded. This facilitated the computation of the exact age (with precision of 2 decimal places). When the latter was unknown (which, fortunately, was rather infrequent), a local events calendar was used to estimate ages.
Measurements were taken using a wooden measuring board, having a precision of
0.1 cm. Children below two years of age were measured lying down. The child was placed in the middle of the board, with the head against the fixed end and legs fully extended by pushing on the knees. The child's soles were at right angle against the movable sliding board.
Children two years of age and above were measured standing up. The child was placed in the middle of the board with heels, calves, buttocks, shoulder blades, and back of head all against the board. The child looked straight ahead while the sliding board was positioned on top of the head crown.
Measurements were made using a 25-kg hanging Salter Scale, having a precision of 0.1 kg. The scale was held by a wooden stick, the child was placed undressed in the weighing pants and the weight was read at eye level when the child was steady and hanged freely. Prior to weighing, the scale was set to zero with the empty weighing pants suspended below.
- Presence of oedema
Oedema is the abnormal accumulation of fluids within extra-cellular spaces. This was detected by applying a moderate thumb pressure on the feet or the front of the legs for three seconds. If there was oedema a shallow print or pit remained when the thumb was lifted. Only children with bilateral oedema (i.e. on both feet/ legs) were considered to have nutritional oedema - an indicator of protein deficiency. When detected, the severity of the oedema was ascertained by assessing the legs and arms.
3. Indicators of nutritional status
- Additional information was collected from mothers on measles vaccination, demography, morbidity, food sources and water and sanitation.
- Classification of surveyed children in terms of their weight for height in percentage of the median was performed on the spot to allow referral to appropriate health facilities if the child was found to be severely wasted or oedematous.
To measure current or acute malnutrition at the time of a survey and detect short-term changes in the nutritional situation, the indicator to be used is weight for height, which expresses the weight of a child in relation to his height. Individual measurements are then compared to reference values for a healthy population (NCHS/WHO standardised reference values10), that is the weight of a child is compared to the distribution of the weights of the reference children of the same height (or length). The weight for height index can be expressed either as standard deviations of the reference distribution, known as Z-scores, or as a percentage of the reference median. Expression in Z-scores has a true statistical meaning, which percentage of the median does not have and is, therefore, recommended.
Global and severe acute malnutrition are defined, as follows:
Global acute malnutrition: proportion of children with W/H index < -2 Z-scores (or 80% of the median) and/or oedema
Severe acute malnutrition: proportion of children with W/H index < -3 Z-scores (or 70% of the median) and/or oedema
Data analysis and calculation of nutritional indices were performed using Epi-Info and Epinut software (version 6.04) and Excel.
- N. Bikin, K. Sullivan, N. Staehling and P. Nierburg. Rapid nutrition surveys. Disasters:1992; 97-103.
- National Center for Health Statistics (1977) NCHS growth curves for children birth - 18 years. United States.