Geographic variation in dengue seroprevalence and force of infection in the urban paediatric population of Indonesia

Abstract

Understanding the heterogeneous nature of dengue transmission is important for prioritizing and guiding the implementation of prevention strategies. However, passive surveillance data in endemic countries are rarely adequately informative. We analyzed data from a cluster-sample, cross-sectional seroprevalence study in 1-18 year-olds to investigate geographic differences in dengue seroprevalence and force of infection in Indonesia. We used catalytic models to estimate the force of infection in each of the 30 randomly selected sub-districts. Based on these estimates, we determined the proportion of sub-districts expected to reach seroprevalence levels of 50%, 70% and 90% by year of age. We used population averaged generalized estimating equation models to investigate individual- and cluster-level determinants of dengue seropositivity. Dengue force of infection varied substantially across Indonesia, ranging from 4.3% to 30.0% between sub-districts. By age nine, 60% of sub-districts are expected to have a seroprevalence ≥70%, rising to 83% by age 11. Higher odds of seropositivity were associated with higher population density (OR = 1.54 per 10-fold rise in population density, 95% CI: 1.03-2.32) and with City (relative to Regency) administrative status (OR = 1.92, 95% CI: 1.32-2.79). Our findings highlight the substantial variation in dengue endemicity within Indonesia and the importance of understanding spatial heterogeneity in dengue transmission intensity for optimal dengue prevention strategies including future implementation of dengue vaccination programmes.

Fig 1. Observed (orange markers) and model-predicted (blue line) seroprevalence by year of age, based on an overall estimated force of infection, λ, of 14.0% per annum, with corresponding 95% confidence intervals, among children aged 1 to 18 years in Indonesia, 2014.

Fig 2. Estimated cluster-specific annual forces of infection with associated 95% confidence intervals among children aged 1 to 18 years in Indonesia, 2014.

Fig 3. Observed (orange markers) and model-predicted (blue line) cluster-specific seroprevalence against estimated cluster-specific force of infection among children aged 1 to 18 years in Indonesia, 2014.

Fig 4. Proportion of clusters expected to reach seroprevalence levels of 50%, 70% and 90% by year of age, based on models with an age-constant force of infection.