Statistical Methods in Public Health

Statistical Methods in Public Health
Chair: Patrick Brown (University of Toronto)
Organizer: Lennon Li (Public Health Ontario)

ROB DEARDON, University of Calgary
Infectious Disease Modelling in the Presence of Underlying Contact Network Uncertainty  [PDF]
Infection disease transmission is often best modelled through contact networks, especially in human populations. These contact networks can represent sexual or social relationships, or shared employment, education or domestic status.

However, contact network data are often difficult and/or costly to collect; there may be privacy issues, or issues of incomplete or incorrect recall in surveys. As a result, contact network data desirable for the modelling may be missing entirely, only partially recorded, or recorded with uncertainty.

Such missing/uncertain data can be incorporated within a data-augmented Bayesian framework using methods such as Markov chain Monte Carlo (MCMC). Here, we consider disease transmission modelling in the presence of uncertain contact networks, touching upon issues of computational difficulty.

ANDREW LAWSON, Medical University of South Carolina
Spatio-temporal Latent Model Choice: With Application in Multiple Disease Small Area Analysis  [PDF]
Space-time variation in disease is the natural focus of many health studies. Conventionally the variation of the space-time risk fields is modeled via random effects. However, it is possible to consider spatial and temporal risk components selected from predefined sets of terms and linked via mixing parameters. In addition there can be important information gained from the analysis of multiple diseases within this context, where shared components could inform about shared environmental risks. In this presentation I will demonstrate the development of multivariate spatio-temporal (MVST) linked models for analysis of shared risk. Some simulation results and a data example based on lung and bronchus cancer and melanoma in South Carolina (1996-2009) will be provided
RHONDA ROSYCHUK, University of Alberta
Cluster Detection Tests for the Surveillance of Emergency Department Presentation Data  [PDF]
Alberta has been one of the few provinces in Canada to have an extensive ambulatory care database that includes both hospital-based and community-based services. Presentations to Emergency Departments are part of the database and when linked with other provincial databases, excellent population-based analyses can result. The provincial health authority sometimes uses this data to provide analyses based on geography. While cluster detection tests have been popular in the statistical literature, they have been underutilized in this jurisdiction. This talk will focus on data issues, cluster detection tests, and knowledge translation activities in the context of Alberta Emergency Department presentation data.