Selection Biases in Repeated Measures Analysis of Treatment Effects for Recurrent Episodic Conditions

• Activity Number: 219
• Type: Invited
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Society for Research on Nicotine and Tobacco
• Author(s): Richard J. Cook*+
• Companies: University of Waterloo
• Keywords: longitudinal data ; selection effects ; inverse weighting

Abstract

Many chronic disease processes feature acute episodic conditions which warrant therapeutic intervention to alleviate symptoms or reduce the risk of further complications. In randomized clinical trials, patients with diseases of this sort are often randomized to one of several treatments and followed over a fixed period, during which any episodes are treated with the assigned treatment. Use of standard methods for longitudinal data is inappropriate however. We consider some particular models to illustrate the biases that result from naive analyses and use data from a recent trial for illustration. We conclude that either fully specified models are required to characterize the process dynamics and render the observation process uninformative, or inverse probability weights must be used to refine generalized estimating equations.