A Bayesian modelling framework for inference of latent infection risk patterns from virus neutralisation assay titration data
Serological assays remain the standard approach for estimating the cumulative incidence of a pathogen and monitoring population immunity. Titration data from virus neutralisation assays are conventionally analysed using a nearly century-old interpolation-based method that neglects inherent imperfections in the assay and produces estimates with no measure of uncertainty. We introduce a two-part Bayesian framework for inferring latent infection risk directly from raw titration data. First, we develop a mechanistic model for serum antibody titration data that estimates antibody concentrations while quantifying uncertainty. Second, we propagate this uncertainty into an age-structured serocatalytic mixture model by integrating over posterior draws of individual antibody concentrations, allowing joint inference on latent serostate membership, force of infection, and serological waning rate. We applied the framework to three cross-sectional serosurveys of the population of England, for enteroviruses A71 (EV-A71) and D68 (EV-D68) and coxsackievirus A6 (CVA6), which are leading causes of severe respiratory illness and hand, foot, and mouth disease. We estimated consistently higher and more persistent lifetime antibody concentrations for EV-D68 than for EV-A71 and CVA6. The proportion of recently infected individuals peaks around 40% by age 5 years for CVA6 and around 32% by age 7 years for EV-A71, before declining with age. These estimates are conditional on a model structure excluding complete seroreversion, which, where identifiable, implied a half-life of several decades. For EV-D68, the inferred proportion previously infected exceeded 50% by age 13 years and grew, with the force of infection declining more gradually with age. These estimates differ from those previously obtained from binarised versions of the same data. Our framework uncovers the wide-ranging variation in antibody levels that are often obscured by conventional endpoint titre methods and infers infection rates directly from raw titration data, without dichotomising at predetermined seropositivity cut-offs, while making minimal assumptions about virus-specific infection mechanisms.