Abstract
This paper presents an integrated statistical approach to the analysis of influenza vaccine lot consistency studies in which three lots are compared. The approach ensures that the overall Type I error rate (i.e., the probability of wrongly concluding that the lots are similar) is controlled. It is argued that the optimum efficacy measure is the geometric mean titer. The approach is demonstrated using data from a randomized, double-blind lot consistency study in which three consecutive production lots of Solvay Pharmaceutical' new, virosomal subunit influenza vaccine Invivac ® were compared.
ACKNOWLEDGMENTS
The author wishes to thank Jeroen Tolboom of the Biometrics department of Solvay Pharmaceuticals, Weesp, the Netherlands, for his valuable comments and suggestions, as well as two anonymous referees.