Importance of data management with statistical analysis set division

Ling WANG,Chan-juan LI,Zhi-wei JIANG,Jie-lai XIA
2015-01-01
Abstract:Testing of hypothesis was affected by statistical analysis set division which was an important data management work before data base lock-in. Objective division of statistical analysis set under blinding was the guarantee of scientific trial conclusion. All the subjects having accepted at least once trial treatment after randomization should be concluded in safety set. Full analysis set should be close to the intention-to-treat as far as possible. Per protocol set division was the most difficult to control in blinded examination because of more subjectivity than the other two. The objectivity of statistical analysis set division must be guaranteed by the accurate raw data, the comprehensive data check and the scientific discussion, all of which were the strict requirement of data management. Proper division of statistical analysis set objectively and scientifically is an important approach to improve the data management quality.
What problem does this paper attempt to address?