areas, and due to the request by MINSA to change areas (As mentioned in the Introduction section.)
As a result the project expanded to 162 communities (46 clusters) . It should be noted that the change in the number of communities (clusters) did not affect the number of beneficiaries. This is due to the fact that the smaller communities included in the original proposal became parts of larger communities, and other communities not originally listed were added accordingly.
The estimates of confidence limits for the survey results were calculated using the following formula:
95% confidence limit of p=pZ (pq)/n
p= proportion/rate in population found from survey
Z= statistical certainty chosen (if 95%, then Z=1.96)
n= sample size
In some of the targeted communities the households are greatly dispersed. Due to this fact, all communities that were selected for the baseline and midterm surveys were chosen again for this final survey. If the village had less than 100 households, the surveys were made in all eligible households until the quota of ten (in some communities 11) was met. If the community had more than 100 households, a list was made and a sampling interval number was selected by dividing the number of households by ten. When the quota was not met, another cluster in the neighborhood was selected as an extension of the first.
The sample consisted of 464 households with children under the age of three years in 162 rural communities. Eligible households were those having at least one living and present child younger than three years of age. It was essential for the written informed consent of the mother for all our procedures, this included blood extraction. If no family members were capable of giving the information, the family was immediately replaced, but this event was very uncommon.
The same methods were used as in the previous surveys to weigh and collect height measurements. For the most part, the children were weighed without any clothing. When clothes were being worn, an amount of 50 to 100g was subtracted to obtain the net weight. Scales (Salter-type, Balper trademark, 100g in precision, 25kg capacity) were adjusted to zero prior to every measurement. Children under