STATISTICS SEMINAR

SPEAKER:

Richard J. Cook

Department of Statistics and Actuarial Science

University of Waterloo

TITLE:

Efficient Robust Methods of Inference for Trials with Recurrent Outcomes Observed over Multiple Treatment Periods

ABSTRACT:

Robust methods have been developed for the analysis of recurrent events based on Poisson estimating functions (Lawless and Nadeau, 1995). Alternative robust methods may be derived based on binomial or multinomial estimating functions when patients' receive different therapies during successive treatment periods. These methods are obtained by suitably conditioning within a "working" class of mixed Poisson models. In many settings such models lead to more efficient estimates and more powerful tests of treatment effects, provided robust variance estimates are adopted. We derive robust pseudo-score tests for testing treatment effects in trials with recurrent event outcomes when baseline periods of observation are available. Adaptations for handling data from cross-over trials are developed. Extensions are then described to facilitate interim monitoring. Simulation studies are reported to assess the frequency properties of the fixed sample and sequential testing schemes and an application is provided for illustrative purposes.

This talk is on joint work with Prof. Wei Wei (University of Michigan) and Prof. Grace Yi (University of Waterloo).

About the Speaker

Richard Cook is a Full Professor in the Department of Statistics and Actuarial Science at the University of Waterloo. His research interests include the development and application of statistical methods to studies of medicine and health. Specific areas of interest include the analysis of event history and longitudinal data, sequential methods, multivariate analysis, clinical trial design, and the assessment of diagnostic tests. Much of his work is in the area of medical statistics and Prof. Cook is a Canadian Institute of Health Research (CIHR) Investigator. Prof. Cook's work is often done jointly with oncologists, rheumatologists and specialists in transfusion medicine. He has also consulted for a number of large pharmaceutical companies. Prof. Cook earned his B.Sc. from McMaster University before going on to complet his M.Sc. and Ph.D. in statistics at the University of Waterloo. As well as his appointment in Statistics and Actuarial Sciences at Waterloo, Prof. Cook currently holds a cross appointment in the Department of Health Studies and Gerentology at Waterloo and also a part-time postion in the Department of Clinical Epidemiology and Biostatistics at McMaster. Among the graduate students that he has superised was the 1997 winner of the Pierre Robillard award for the best statistics PhD thesis.

References

You can download this talk in pdf format or postscript format.

Some relevant background references are

• Andersen, P. K. and Gill, R. D. (1982). Cox's regression model for counting processes: a large sample study. The Annals of Statistics, 10, 1100-1120.
• Cook, R. J., Lawless, J. F. and Nadeau, C. (1996). Robust Tests for Treatment Comparisons Based on Recurrent Event Responses. Biometrics, 52, 557-571.
• Farewell, V. T. and Sprott, D. A. (1988). The use of a mixture model in the analysis of count data. Biometrics, 44, 1191-1194.
• Lawless, J.F. and Nadeau, J.C. (1995). Some simple robust methods for the analysis of recurrent events. Technometrics, 37, 158-168.

Department of Mathematics and Statistics

Graduate Program in Statistics

This page is maintained by Angelo Canty,
Last updated on October 8, 2002