Sample Size Calculations for Repeated Binary Population Pharmacodynamic Experiments
Kayode Ogungbenro and Leon Aarons
Centre for Applied Pharmacokinetics Research, 3School of Pharmacy and Pharmaceutical Sciences, The University of Manchester, Oxford Road, Manchester, M13 9PT, United Kingdom
Objectives: Repeated binary measurements have been analysed using mixed effects modelling techniques and there is no analytical method for calculating sample size for hypothesis testing. The aim of this work was to develop an analytical method for sample size calculations for repeated binary measurements based on analysis by mixed effects modelling.
Methods: The method proposed by Rochon (1) for GLM based on analysis by GEE has been extended to repeated binary measurements under mixed effects modelling and using a logit transformation. The method can be used for calculating the sample size required to detect a difference in a parameter between different groups, say placebo and drug. Wald's test has been used for hypothesis testing. Extensions to account for unequal allocation of subjects across groups and unbalanced sampling designs between and within groups have also been described.
Results: The method described has been used to calculate the sample size required to detect a difference in the slope parameter between two groups in a linear model for repeated binary measurements. Equal and unequal sample sizes as well as balanced and unbalanced sampling designs were considered. The results showed good agreement between nominal power and the power obtained by simulations in NONMEM. The results also showed that for balanced sampling designs sample size increases with increased variability only on the slope parameter and for unbalanced sampling designs, sample size increases with increased variability on intercept and slope parameters. The results also showed that designs that involve sampling based on an optimal design require a reduced sample size.
Conclusions: A fast and efficient method for calculating sample size for repeated binary population pharmacodynamic method has been described. A carefully designed trial will produce an efficient trial and can help to reduce cost.
References:
[1] J.Rochon. Application of GEE procedures for sample size calculations in repeated measures experiments. Stat.Med. 17:1643-1658 (1998).
