Two Independent Prospectively Planned Blinded Weibull Statistical Analyses of Flexural Strength Data of Zirconia Materials

摘要

Zirconia as a restoration dental material are gaining attention because of their high mechanical properties and good biocompatibility. Therefore, investigation of the flexural strength of zirconia is of great interest. For this purpose, Weibull statistics for description of the material reliability are frequently used. The aim of this work was to present a blinded data set to two independent statisticians for two parallel statistical analyses in order to find an optimal statistical approach for analysis of in-vitro measured flexural strength data of zirconia materials. A prospectively planned independent blinded statistical analysis implementing three quality control actions “blinded data set”, “independent statistical analyses” and “parallel manuscript writing” was designed. Statistical analysis paths taken by both biostatisticians differed. They arrived at complementary results. The major difference was caused by two alternative distributional assumptions (Weibull/Normal) and alternative fitting methods (LS/ML). The parallel statistical analysis and manuscript writing approach on a blinded data set greatly supported our choice of statistical methods for analysis of flexural strength results of zirconia materials.