The risk approach in diarrheal disease intervention in Grenada

In a previous paper (Soc. Sci. Med. 23, 995-1002, 1986), we described the development of a method for identifying households at high risk of childhood diarrhea based on a logit analysis of data collected as part of a diarrhea disease control project in Grenada. In this paper, we report the results of a follow-up study designed to replicate the first using data on diarrhea incidence and risk factors collected in a different set of rural communities in Grenada. The findings of the first study were not replicated. There was no significant association between the risk factors and diarrhea in the follow-up study in contrast to the very strong associations found in the first study. As a consequence, it was not possible to construct a logit model from the second data set. The possible reasons for the contrasting sets of results and their implications for the application of the risk approach are discussed with reference to a two by two true-false table.