Applications of random effects Poisson regression model for repeated count data in clinical trial

Objective To discuss random effects Poisson regression model forrepeated count data in clinical trial. Methods The advantage and disadvantage werecompared for fixed and random Poisson regression models when the subjects weremeasured repeatedly through time. An example of assessing clinical equivalencewas illustrated. Results The within-subject correlation and information measuredrepeated through time were both considered if random Poisson regression wasused. The type I error arise by fitting a fixed model for each time point separately,and the type II error arise by fitting a fixed model for all the data without considerthe within-subject correlation, would be deduced by fitting a random model.Conclusion The random effects Poisson regression model is valid, effective forrepeated count data in clinical trial, and more flexible in dealing with random missingdata.