Multivariate Spatial Disease Surveillance Using a Latent Shared Component Model

• Activity Number: 597
• Type: Topic Contributed
• Date/Time: Thursday, August 2, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Ana Corberan-Vallet*+ and Andrew B Lawson
• Companies: University of Valencia and Medical University of South Carolina
• Keywords: Public health surveillance ; multivariate spatial data ; shared component model ; conditional predictive ordinate

Abstract

Effective disease surveillance is fundamental to facilitate timely public health response and, consequently, to reduce morbidity and mortality. Recent developments for the analysis of disease data in a univariate surveillance setting use a statistical model to describe the endemic behavior of disease and the emphasis is placed on detection of unusual departures from predictable patterns based on the estimated model. In this work, we develop a Bayesian model-based surveillance technique for the analysis of spatial data on multiple diseases which are likely to be correlated. In particular, we present a new formulation for the shared component model to describe the behavior of diseases under non-epidemic conditions. A novelty of our formulation is the use of indicator variables, which allow us to identify shared and disease-specific latent spatial fields describing the risk surface for each disease. We show then how the conditional predictive ordinate can be adapted in a multivariate surveillance setting to detect areas of increased disease incidence.