LQAS in Health Monitoring – Insights from a Bayesian Perspective

Abstract

Lot Quality Assurance Sampling (LQAS) is strongly advocated for use in monitoring the health status of populations, largely in the developing world. It is advocated both for the monitoring of small areas as well as for making global assessments of the health status of a larger region. This paper contrasts the interpretation offered by LQAS methods to that offered by Bayesian hierarchical models. It considers applications to previously reported local area data and presents a reanalysis of published data on vaccine coverage in Peru as well as HTLV-1 prevalence in Benin. The desirability of using Bayesian methods in the field may be challenged; nevertheless this work amplifies previously expressed concerns about the way the LQAS method can be used. It raises questions about the ability of the LQAS approach to make, sufficiently often, the correct decisions in order to be useful in monitoring health programmes at the local level