Sampling Methods: WHO-EPI 30-Cluster, Systematic Random Sampling, and Segmentation

Measuring vaccination coverage permits evaluation and appropriate targeting of vaccination services. Population-based cross-sectional surveys should be as representative as possible, within realistic cost parameters, to determine the true percentage of children within a certain geographic area that have been vaccinated. Numerous sampling methods are available for such surveys, each with their own strengths and limitations.

3 sampling methods will be compared: 1) The WHO-EPI 30-Cluster survey was popularized by WHO in the 1970s. It is feasible in a wide variety of settings; however, household selection is biased towards the center, it is not a true probability survey, and all households are selected from one area within the cluster. 2) Systematic random sampling does not bias towards the center, and spreads household selection over the entire cluster; however, considerable walking is required because interviewers must canvas the entire cluster 3) Segmentation is unbiased and requires less walking; but households are selected from one area within the cluster. This presentation will describe the methods for conducting each of these sampling options, their pros and cons, and situations in which each is recommended. In addition, important “non-sampling” issues that should be considered when conducting a coverage survey will be discussed.