A review of the limitations of statistical analysis of treatment outcomes used in periodontal research

Abstract

Results of clinical trials are reported in terms of statistical significance, but interpretation of statistical significance in relation to clinical benefits is limited. Furthermore, the compromises inherent in the design of clinical trials, and in the statistical analysis techniques themselves cast doubts on the soundness of the inferences drawn from data from clinical trials. The aim of this paper is to review the published research on data management and interpretation in periodontal research, and to attempt to find ways in which the results of clinical trials could be reported in a more meaningful way, in terms of treatment outcomes. It is concluded that due to the limitations of current statistical methods and trial design, that clinical outcome variables should be reported in addition to statistical significance. Furthermore, due to the multiplicity of sites, subjects, and variables used to report on clinical conditions that may not be linear in their progression or regression, care needs to be exercised in interpretation of the results of clinical trials to avoid reporting positive outcome results that arise out of the limitations of current statistical analysis techniques or computational errors rather than clinical changes.