Quantifying the Unseen Dengue Burden: A Municipality-Level Ascertainment Surface for Pará State, Brazilian Amazon

Passive dengue surveillance systematically underestimates true disease burden, and this under-ascertainment is spatially structured rather than random. In the Brazilian Amazon, municipalities with the weakest notification infrastructure tend to coincide with those presenting the most favorable environmental conditions for vector proliferation. This work investigates whether municipalities with more active surveillance can serve as a structural proxy to infer cases in less active regions, allowing for the identification of localities where notification capacity fails to reflect true transmission. Building on this framework, this study aimed to estimate municipality-level dengue under-ascertainment in Pará by disentangling surveillance capacity from true transmission intensity through a Bayesian spatiotemporal model. A BYM2-ST model was fitted to dengue notification data from municipalities of Pará, from January 2018 to December 2024, integrating socioeconomic indicators and time-varying climatic covariates. A spatial ascertainment ratio was derived for each municipality from the model, benchmarking observed notifications against the structurally expected counterfactual. The results showed that only 20.1% of municipalities reported notification counts within the expected range, while 40.3% showed statistically supported evidence of under-ascertainment.