Workshops and conferences >

Workshops:

• Record linkage in studies of population health — an overview
  
  
• Methods for analyzing longitudinal health survey data - Theory and applications

Record linkage in studies of population health — an overview

Karla Fox, Department of National Defense and Patricia Whitridge, Elections Canada (French and English presentation with simultaneous translation and with English and French materials)

Record linkage is simply the integration of information from two or more independent sources. Under this framework, records are linked on the basis of common data. Record linkage has become increasingly important in many different domains: maintenance of electronic registry information, health care administration, demographic studies and medical research. Whether following a cohort over time or linking patients to vital statistics in order to calculate survival curves, researchers need to understand record linkage.

Using examples from different applications in the health field, this workshop will cover the general principles of record linkage (statistical matching and exact matching), including data preparation, linkage techniques and linkage evaluation. Practical examples will be used to illustrate different points throughout the course. Participants will gain an appreciation for the different concepts and principles involved and an understanding of how to approach a problem requiring record linkage. At the end, students will be provided with a comprehensive list of references in the field of record linkage.

Methods for analyzing longitudinal health survey data - Theory and applications

Prof. Mary Thompson, University of Waterloo, Waterloo, Canada (English presentation with simultaneous translation and with French and English materials)

The workshop will begin with an overview of models for longitudinal data, with examples from the health sciences. Repeated measures models considered will include growth curve, GEE, and stochastic process models. Event history models will include survival and recurrent models. Types of explanatory variate (fixed, time dependent, internal, external) will be reviewed, as well as types of missing data. The overview will be followed by a discussion of observational studies and causality, through examples illustrating the use of longitudinal or quasi-longitudinal data in examining causal hypotheses. The role of path diagrams in modeling dependencies will be considered. Principles of adapting the models and methods to complex survey data will be outlined, with examples from Canadian health survey data. Problems presented by models with complex likelihoods (particularly those arising from missingness or latency) will be illustrated. The workshop will end with a discussion and limited environmental scan of software, available or known to be under development, for longitudinal health survey data.