As described earlier, we define a study area or cluster as the collection of villages and/or settlements that fall within the catchment of health clinic/s or distribution site/s that is/are currently implementing MAM treatment and will eventually implement the MAM prevention package/s. We recommend that an area or cluster should have at least 24 villages or settlements within its catchment to allow for sufficient sample sizes to be selected in each one. This may mean that for the purposes of the study, some adjacent areas or clusters maybe joined together to form a single study area or cluster in order to reach the minimum number of villages or settlements within it.
As already mentioned in the outcome measures section, we recommend selecting areas or clusters that have existing MAM treatment (TSFP) as this will be the ‘baseline’ or ‘control’ status. In addition to this, we would want selected study areas or clusters be roughly of the same ‘baseline’ status with regard to other nutrition or nutrition-related intervention such as IMCI, etc.
The issue of contamination between study areas or clusters that are still in ‘baseline’ status and those that have switched over to intervention should also be considered during selection. As much as possible, selection should be done such that the risk of contamination is minimised. This would mean selecting study areas or clusters that are non-adjacent to each other.
Finally, study area or cluster selection should take into account as well the level of data collection intensity required for this study. Good and reliable frequent access to the study areas or clusters is necessary in order to complete the study.
Given these considerations and based on what we know about the different states in Sudan, we think Kassala state would be the most suitable location for the study.
Ideally, the schedule in which a cluster switches over to becoming an intervention should be randomised which would require us to have influence on the rollout plan of the interventions in the 12 selected clusters for the study. If this is possible, after the 12 study areas or clusters are selected, we will randomly determine the order by which each cluster switches over to intervention during the one year period of the study.