|
Activity Number:
|
135
- Multiplicity, Missing Data and Other Topics
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 9, 2021 : 1:30 PM to 3:20 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
|
Abstract #317952
|
|
|
Title:
|
Statistical Methods to Assess In-Vitro Comparability
|
|
Author(s):
|
Michael Daniel Cid Lucagbo* and Tianjiao Dai and Yixin Ren and Meiyu Shen and Yi Tsong
|
|
Companies:
|
University of Maryland, Baltimore County and University of the Philippines Diliman and U.S. Food and Drug Administration and Merck & Co., Inc. and Office of Biostatistics CDER, FDA and CDER, US FDA
|
|
Keywords:
|
Comparability;
quality range;
tolerance interval
|
|
Abstract:
|
To assess the comparability of test and reference products, Mielke et al. (2019) proposed a statistical hypothesis testing procedure based on a quality range approach under the normality assumption of the quality measurements. They proposed to test the probability that the quality measures of the test product fall outside the quality range of the reference product. The test product is claimed to be comparable to the reference product if the difference in the probabilities outside the range determined by the reference quantiles is no more than a pre-specified comparable margin. There are statistical issues with the approach of Mielke et al. (2019). One of them is that the quality range is assessed by the asymptotic method, which leads to an underestimate of the range and the over-conservative method of the test. In this talk, we discuss alternative approaches to determine the quality range. These approaches include correcting the reference interval coverage so that the resulting interval gives 95% confidence and computing the reference quality range as tolerance limits. The proposed approaches can control the Type I error rate and have asymptotic powers of 1.
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2021 program
|
