Estimating Polio Campaign Coverage and Immunity Status from Dose Histories

• Activity Number: 554
• Type: Contributed
• Date/Time: Wednesday, August 12, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Arend Voorman* and Alexander Upfill-Brown and Hil Lyons and Guillaume Chabot-Couture
• Companies: Bill and Melinda Gates Foundation and Institute for Disease Modeling and Institute for Disease Modeling and Institute for Disease Modeling
• Keywords: bayesian statistics ; small area estimation ; spatial epidemiology

Abstract

The success of the Global Polio Eradication Initiative depends critically on the quality of house-to-house polio vaccination campaigns across Africa and Asia. However, administrative coverage data is often unreliable and post-campaign coverage surveys routinely over-estimate coverage due to sampling bias. We present a new model for estimating campaign coverage based on reported polio vaccine doses of non-polio paralysis cases, which can be used to identify areas of poor coverage and to plan appropriate campaigns to interrupt or prevent disease transmission. We use a Bayesian regression approach, where doses reported are regressed on campaign exposure in order to estimate time-varying coverage; we then correct for demographic bias in estimating the immunity status of the population. To handle sparse data, we smooth campaign coverage over space and time. We illustrate our model using data from Nigeria, Somalia, and Pakistan, and validate the results by comparing predicted population immunity to the location of confirmed poliovirus cases. We also investigate extensions of the model which allow coverage to vary by observed demographic variables or latent class.