Spatial Model for Risk Prediction and Subnational Prioritization to Aid Poliovirus Eradication in Pakistan

• Activity Number: 130
• Type: Contributed
• Date/Time: Monday, August 1, 2016 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Laina Mercer* and Steve Kroiss and Hil Lyons and Guillaume Chabot-Couture
• Companies: Institute for Disease Modeling and Institute for Disease Modeling and Institute for Disease Modeling and Institute for Disease Modeling
• Keywords: Poliovirus ; Spatiotemporal smoothing ; Poisson hurdle ; Disease eradication ; Hierarchical Bayesian Models

Abstract

Pakistan is one of only two countries where poliovirus circulation remains endemic. For the Pakistan polio eradication program, identifying high risk districts is essential to target interventions and allocate limited resources. Using a hierarchical Bayesian framework we developed a spatial Poisson hurdle model to jointly model the probability of one or more cases, and the number of cases which would be detected in the event of an outbreak. In the development of this model, we used spatiotemporal smoothing methods to generate sub-national estimates of likely predictors of poliovirus cases, such as routine immunization rates. This smoothing methodology was applied to vaccine dose histories and poliovirus case information from Pakistan's acute flaccid paralysis (AFP) surveillance data, from 2003 to 2015. For our application, we compared various spatial hierarchical models using a variety of measures, including the area under the curve (AUC) based on ranked risk. The results of this model have been used to inform which sub-national areas in Pakistan should receive additional immunization activities, additional monitoring, and other special interventions.