|
Abstract:
|
In clinical trials, effectiveness is usually assessed using responder analysis, where continuous measures of disease activity are dichotomized as responder or non-responder. Dichotomization can greatly simplify the statistical analysis with easy interpretation and presentation of results. However, it is well known that such dichotomization generally causes some serious drawbacks, for example, loss of information, which leads to loss of statistical power, underestimate of the extent of variation in outcome measures between responder and non-responder groups, and ignorance of potential non-linear relationships between variable(s) and outcome measures. In contrast, such issues can be avoided if analyses of continuous measures can be conducted. The purpose of this paper is to compare statistical analyses on both dichotomized endpoints and continuous endpoints in vaccine trials and to discuss the advantages of utilizing continuous measures, geometric mean of continuous measures, and adjusted geometric mean of continuous measures in estimating antibody response and vaccine immunogenicity.
|
