Estimation of Sufficient Core Groups for Trachoma Transmission in Heterogeneous Populations Using a Novel Agent-Based Stochastic Model

Abstract

Purpose : Mass drug administration has practical limitations and has proven ineffective in eliminating trachoma from some endemic populations. In these cases, antibiotic treatment targeted at the most transmissive subset of the population may prove to be a more effective elimination strategy. We developed a novel agent-based stochastic SIS model of trachoma transmission that accounts for heterogeneity in trachoma transmissibility and susceptibility amongst individuals and applied graph theory principles in an attempt to derive a more precise estimate of the minimal core group, the smallest set of individuals whose removal results in elimination of disease from a population.

Methods : A 100-node network of agents was constructed. Instantaneous transmission and recovery probabilities (represented in the graph as directed edges and self-loops, respectively), were calculated for all agents based on known β,γ,ε and up to 3 individual parameters H,TI,O known to increase transmissibility (H: agents in the same household, TI: intense trachoma inflammation, O: ocular discharge). Parameters were assigned to each agent in accordance with a random sample from an anonymized dataset of 312 Ethiopian children of 0-5 years of age enrolled in a multi-year trachoma study. Core groups were found by consecutively removing agents with the highest out-degree centrality until the mean equilibrium prevalence, as determined by 250 Monte Carlo simulations, fell to < 0.05.

Results : The specific agents comprising the MCG were successfully identified for different combinations of parameters. The sizes of the MCGs for the parameter combinations were as follows: Homogeneous (no parameters): 42; H: 42; H+TI: 42; and H+TI+O: 45. The successive addition of transmission risk factors progressively refined the MCG over the homogeneous MCG baseline.

Conclusions : A novel agent-based model of trachoma transmission was used to refine MCG estimations for a heterogeneous test population. This model provides a more generalizable framework for examining the effects of varying transmission and susceptibility parameters, prevalence rates, and MCG determination algorithms on core group size, which may assist in devising optimal strategies for trachoma elimination and control.

This is a 2020 ARVO Annual Meeting abstract.

 

Logistic curve behavior of trachoma prevalence converging to equilibrium in a Monte Carlo transmission simulation with 3 parameters (H, TI, O)