2018 ESA Annual Meeting (August 5 -- 10)

OOS 32-4 - Integrating clinical, cross-sectional, and time series data to infer cross-scale disease dynamics and inform control

Thursday, August 9, 2018: 2:30 PM
345, New Orleans Ernest N. Morial Convention Center
Micaela E. Martinez, Environmental Health Sciences, Columbia University, New York, NY
Background/Question/Methods

Infectious disease eradication is anthropogenically-driven extinction of a pathogen species. There are over a dozen licensed vaccines, with the potential to be used for disease eradication. To best leverage existing vaccines, and make informed policy decisions, vaccine mode of action and transmission impact must be accurately quantified. Vaccine mode of action is the manner in which a vaccine confers protection (e.g., by reducing susceptibility to infection and/or reducing the pathological consequences of infection). The mode of action determines the transmission impact, which we define as the realized transmission reduction at the population-level, per capita immunization. The amount of immunization required in a population to achieve herd immunity, and locally eradicate infection, scales with the transmission impact. As with other ecological systems, host-pathogen systems display variation in space and time. Due to variation in climate, host demography, and other factors that shape transmission, the transmission impact afforded by a vaccine may differ among ecological contexts. Here we address the question: how can vaccine mode of action and transmission impact be evaluated in real-world and/or real-time settings? Using measles, varicella (chickenpox), and polio as study systems, here we demonstrate a generalizable framework for integrating clinical data, cross-sectional serosurvey data, and time series data (cases and demography) to infer vaccine mode of action and transmission impact.

Results/Conclusions

We show that traditional Susceptible-Infected-Recovered (SIR) models can be modified to facilitate the integration of (1) clinical data on vaccine responses, (2) serology data, particularly serology detailing maternal immunity in infants, and time series data on (3) notifiable diseases cases, and (4) host demography. Using a combination of simulation studies and statistical inference (i.e., maximization by iterated particle filtering) we were able to integrate clinical and serological data to demonstrate that the dose timing of measles-mumps-rubella vaccine can be changed to reduce susceptibility in infants, while maintaining herd immunity. We were also able to delineate differences in vaccine mode of action and transmission impact of the two polio vaccines, IPV and OPV. Lastly, we will share preliminary results for "puzzle-piecing” data within models to infer the transmission impact of varicella vaccination, which is heterogeneous and debated in its use worldwide.