Influence of criteria on the results of in vitro evaluation of microleakage

Objectives. The aim of this study was to compare and explain the statistical methods employed to evaluate the in vitro sealing efficiency of adhesive restorative systems. Methods. Two hundred and sixty sound freshly extracted human premolars were randomly divided into 13 groups. Standardized cavities were prepared, and the teeth were restored with 13 restorative systems. The teeth were thermocycled, immersed in dye, embedded in resin and sectioned. Five evaluation criteria were recorded: mean, median and mode of the data measured on each tooth, maximum dye penetration measurements on each tooth, and percentage of teeth in each group without any dye penetration. For each parameter, one-way ANOVAs and Duncan a posteriori tests were used to compare the 13 systems. Results. The number of non-statistically different subgroups, pointed out by Duncan tests, was greater when the selected criterion was the maximum dye penetration (6 subgroups) or the percentage of teeth without any penetration (5 subgroups) than when the criterion was the median (3 subgroups), the mode or the mean (4 subgroups). The positioning of the 13 adhesive restorative systems established from the five criteria was different. Equivalent adhesion strategies revealing different experimental results indicate that other factors contribute to the final effectiveness of a particular system: clinical approach with respect to the formation of an elastic bonding layer, and shrinkage, physical and rheological properties of resin composite. Significance: The results of these in vitro studies of dye penetration must be considered as comparisons and not as absolute conclusions. The maximum dye penetration measured on each tooth, which complies with the aim of the in vitro evaluation of sealing efficiency defined by Pashley (1990) and allows powerful statistical analysis of results, seems to be the best evaluation criterion.