Evaluation of test statistics in split-mouth clinical trials.

This study was undertaken to examine the appropriateness of test statistics used for split-mouth clinical trials. Twenty-two published trials were reviewed and the primary test statistics were labeled as appropriate, inappropriate, or absent. Only 5 of the 22 trials reported an appropriate test statistics. Of the remaining 17 (rials, 12 used an inappropriate lest statistics, and 5 did not report or did not use statistical tests. A serious problem is that more than half of the reviewed trials have reported the use of a one-way analysis of variance or a two-sample t test to investigate the data of the split-mouth trial. This mistake may have led to a large increase of the Type II errors due to the correlated nature of split-mouth data. Failure to detect genuine therapeutic differences because of inadequate attention to the data analysis can occur with inappropriate use of test statistics. It is concluded that there should be more concern for the use of appropriate test statistics. Accepted for pubiication September 27, 1988