The distribution of trachoma across communities in Malawi, Mozambique, and Tanzania characterizes progress towards elimination and the probability of transmission hotspots

Abstract

Purpose : The WHO has targeted the control of trachoma by 2030. Control is defined by having <5% clinical activity (follicular trachoma, TF) in all endemic districts. However, TF can persist for years even after the chlamydial infection that causes the disease is gone. Repeated surveys and treatments of areas with lingering TF can waste resources better allocated for true hotspots. Theoretical models suggest that if trachoma is disappearing in a district, the distribution across communities within a district would follow a geometric distribution.

Methods : We assessed the prevalence of TF in 3 trachoma-endemic countries: Malawi, Mozambique, and Tanzania. Data included trachoma surveillance surveys (TSS) administered at least 24 months after stopping interventions. The countries' TSS were divided into districts, and communities within each district. We used goodness-of-fit testing to assess whether each district’s community-level surveys arose from a geometric distribution and estimated a False Discovery Rate (FDR) representing the probability that a district is controlled. We also used a Bayesian Hierarchical Model to estimate the probability that any district was a hot spot. To contrast the TSS, we conducted a similar analysis with Baseline (BL) and Trachoma Impact Surveys (TIS).

Results : The total number of TSS communities and districts included in this study were: Malawi (1260 in 47 districts), Mozambique (1032 communities in 43 districts), and Tanzania (1970 communities in 75 districts). The median district-level TF prevalence for all three countries was 1.6% with an interquartile range (IQR 1.0%--2.7%). The percentage of districts and villages with TF> 5% was 9.7% and 14.3%, respectively. No TSS district had a False Discovery Rate <18%, suggesting no strong evidence that any were hot spots. Some BL and TIS districts were inconsistent with a geometric distribution, and we estimated probabilities of being a hotspot, not controlled without further intervention.

Conclusions : The distribution of TF prevalence measurements in areas where interventions have stopped was always consistent with a disappearing disease. On the other hand, some surveys of districts where interventions are still taking place suggested existence of transmission hot spots that require enhanced interventions for control.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.