Abstract:
|
In clinical trials, count data often have a variance greater than the mean, which is called overdispersion. For overdispersed count data, the ordinary Poisson model cannot be used due to the assumption of equi-dispersion. The generalized Poisson regression model is used as one of approaches to model overdispersed count data. In this presentation, the sample size formula from the Wald test is derived for the generalized Poisson regression model comparing rates of recurrent events with equal or unequal follow-up time. The performance of the derived sample size formula is assessed using simulations with various settings. The simulation study indicates the formula is appropriate for overdispersed count data with the generalized Poisson distribution.
|