Assessment of P values for demographic data in randomized controlled trials

Abstract

In a large number of randomized controlled trials, researchers provide P values for demographic data, which are commonly reported in table 1 of the article for the purpose of emphasizing the lack of differences between or among groups. As such, the authors intend to demonstrate that statistically insignificant P values in the demographic data confirm that group randomization was adequately performed. However, statistically insignificant P values do not necessarily reflect successful randomization. It is more important to rigorously establish a plan for statistical analysis during the design and planning stage of the study, and to consider whether any of the variables included in the demographic data could potentially affect the research results. If a researcher rigorously designed and planned a study, and performed it accordingly, the conclusions drawn from the results would not be influenced by P values, regardless of whether they were significant. In contrasts, imbalanced variables could affect the results after variance controlling, even though whole study process are well planned and executed. In this situation, the researcher can provide results with both the initial method and a second stage of analysis including such variables. Otherwise, for brief conclusions, it would be pointless to report P values in a table simply listing baseline data of the participants.

Keywords

• baseline
• bias
• characteristics
• demographic data
• difference
• p value
• randomization
• randomized controlled trial
• variable