Simulation study to assess the precision and accuracy of the two‐stage cluster survey for injection safety

Abstract

A simulation study was conducted to assess the accuracy and precision of the proposed sampling design for the World Health Organization (WHO) Injection Safety Assessment Survey. The proposed sampling design specifies that 8 clusters would be selected with probability proportionate to the size (PPS) of the population, and within each of the selected clusters, 10 health care facilities would be randomly selected. Two artificial populations were created based on the two African countries in which the assessment was pilot tested. To create a wide variety of hypothetical populations, the assignment of whether a health care facility was safe or not was based on the different combinations of the population proportion of safe health care facilities in the country and the homogeneity of the districts in the country with respect to injection safety. Based on results obtained through univariate and multi‐factor analysis of variance, it was found that the absolute bias, standard error, and MSE are all significantly smaller when the true population proportion of safe health care facilities for both the urban and rural clusters is small (p=0.10) or large (p=0.90), as compared to the populations with differing urban and rural proportions; and that the populations with homogeneous clusters (i.e., standard deviation =0.01) had lower absolute bias, standard error, and MSE. From this study, we conclude that this sampling design yields an unbiased estimate of the true proportion of safe health care facilities in a country, and has an average precision of plus/minus 13.6 percentage points.