Design parameters for population-based trachoma prevalence surveys

Overview

Trachoma results from infection with particular strains of the bacterium Chlamydia trachomatis, causing blindness in the world’s poorest people. In 1996, a World Health Organization (WHO) Alliance was established to support elimination of the disease as a public health problem1 by 2020. Decisions on where and how to implement the “SAFE” (surgery, antibiotics, facial cleanliness, environmental improvement) strategy for elimination and on whether or not elimination has been achieved rely on estimates of the prevalence of disease.

The gold-standard approach for estimating disease prevalence is a population-based prevalence survey (PBPS), adequately powered for the disease of interest. In an effort to stretch scarce resources, since 1996, various cheaper and cruder methods have been designed to assess the burden of trachoma and the potential need for interventions against it, including trachoma rapid assessment, acceptance sampling trachoma rapid assessment and integrated threshold mapping. Each of these methods has epidemiological drawbacks accompanying its lower cost. Robust prevalence estimates are important for programmes wishing to establish whether mass drug administration of antibiotics should begin [at baseline survey] or could be safely discontinued [at impact survey], or to determine whether the disease has recrudesced beyond elimination thresholds after cessation of antibiotic mass drug administration [at pre-validation surveillance survey]. PBPSs are used to generate such estimates and are usually performed through cluster sampling.

Baseline surveys have now been mostly completed in suspected trachoma-endemic populations worldwide using a highly standardized PBPS approach consistent with recommendations previously published by WHO. The current document provides WHO recommendations, with justification, for undertaking impact and surveillance surveys. The unprecedented recent expansion of the global trachoma programme, which anticipates a parallel, trailing acceleration in demand for impact and surveillance surveys, warrants urgent dissemination of these recommendations.

In developing these recommendations, several resources have been employed. First, existing WHO guidance on trachoma prevalence surveys was reviewed using electronic and manual searches of WHO publications on trachoma located, respectively, on the WHO website and in the personal collections of those preparing this document. Second, basic statistical principles were applied to calculate sample size requirements for various epidemiological scenarios.

To help parameterize the sample size calculations, use was made of the survey experience acquired from 2012 to 2016 within the Global Trachoma Mapping Project (GTMP) and the Tropical Data service, which has supported national programmes to complete trachoma prevalence surveys since the completion of the GTMP.