|
Abstract:
|
It is still common to discard data if a subject did not finish a sequence in clinical pharmacology studies. The analysis of data from an unbalanced, incomplete cross-over design in SAS procedures may not include all available data to make maximum use of the data. These kind of studies are relatively small. The more data are used in the analysis, the better. If the FDA's guidance for industry, Statistical Approaches to Establishing Bioequivalence, is followed closely, subjects who did not complete the study sequences are automatically excluded from the analysis. For this reason, the subjects also are excluded from the summary statistics in many study reports. With the advance of general linear model research in the past decade, is this still a good practice?
|
