|
Abstract:
|
Combining potency estimates is a necessary step when multiple bioassays are carried out to estimate the potency of a test preparation. Early references, Finney (1978) and Bliss (1952), provided a two-step procedure testing homogeneity first followed by a weighted combination. Later the USP and EP compendia provided both weighted and unweighted methods for combining estimates. All of these methods assume at least conditional independence among the estimates. Modern bioassay analytical procedures can use designs which violate the independence assumption. Consequently, appropriate weighting must be used in the analysis to produce unbiased estimates of uncertainty. The nature of this weighting as a topic deserves wider attention. The round table participants are invited to share their experiences in applying the various methods of combining potency estimates, assessing the relative pros and cons of the various methods from a statistical and regulatory perspective. Accommodating correlations will be a key question for discussion.
|
