A new endpoint definition improved clinical relevance and statistical power in a vaccine trial

作者：Pedrono Gaelle; Thiebaut Rodolphe; Alioum Ahmadou; Lesprit Philippe; Fritzell Bernard; Levy Yves; Chene Genevieve^*

来源：Journal of Clinical Epidemiology, 2009, 62(10): 1054-1061.

DOI：10.1016/j.jclinepi.2008.10.014

• Vaccine
• endpoint
• proportional odds model
• HIV infection
• statistical power
• random effect

Objective: Endpoints used for the evaluation of immunogenicity in vaccine trials are often the proportion of individuals with immune response or geometric means of antibody concentrations for each serotype. When a vaccine includes several types of the same species, we illustrate how an endpoint combining all responses may improve clinical relevance and statistical power.
Study Design and Settings: The motivating example was the ANRS 114 Pneumovac trial where the effect of two vaccine strategies against Streptococcus pneumoniae was assessed in adults infected by the Human Immunodeficiency Virus. The power associated with several endpoints was calculated in the example and in simulations. A new endpoint based on four ordered levels is formulated and analyzed by using a proportional odds model.
Results and Conclusion: The analysis of this new endpoint led to an odds ratio allowing detection of improvement and detriment. In the simulation study, this endpoint was associated with the largest statistical power by increasing the amount of information used as compared with usual endpoints. We recommend this new endpoint formulation in the formal development of a new vaccination regimen, whenever applicable.

• 出版日期2009-10