Edi Prianto

Targeted dengue interventions require reliable estimates of transmission intensity and population immunity at the local level. The force of infection (FOI) provides an objective measure of transmission intensity, but its estimation traditionally relies on resource-intensive seroprevalence surveys. We developed a hierarchical extension of existing catalytic models to estimate FOI using routine age-stratified surveillance data, allowing partial pooling of information across districts within provinces. We applied this approach to dengue surveillance data from Jakarta and West Java provinces, Indonesia, and compared it with non-hierarchical implementations. Both hierarchical and non-hierarchical approaches produced FOI estimates consistent with 2014 seroprevalence data. The hierarchical framework provided more robust estimates through partial pooling under varied data availability scenarios but showed sensitivity to age-stratification choices and could miss district-specific patterns when local epidemiology differed from regional trends. Model comparison using expected log pointwise predictive density showed that accounting for overdispersion through negative binomial likelihood improved model performance regardless of hierarchical structure. Our analysis showed distinct patterns in reporting parameters between provinces, with Jakarta showing higher reporting rates despite lower FOI estimates than West Java. Implementation of the hierarchical framework requires understanding of local dengue epidemiology, as clustering districts with different epidemiological profiles could produce inaccurate estimates.