Cluster sampling for immunization coverage.

The WHO/EPI cluster sampling method for immunization coverage surveys is part of the course for management training in EPI programmes. The application of this method, based on a framework of villages, is impractical in dispersed populations common in many countries in Africa. To make the method work under those circumstances leads to unacceptable bias, which tends to overestimate the real coverage rate. A modified random cluster sampling method for dispersed populations is presented.