Meta-analytical review of factors involved in dentin adherence.

Literature data on adherence tests of dentin-bonding systems (DBS) may differ widely, even for the same DBS. The problem of bond testing is that materials are seldom compared with a standard, and experimental conditions often vary. We sought to identify the parameters that influence this variability. Using inclusion and exclusion criteria, we conducted a meta-analytical review of 75 articles, published between 1992 and 1996 in SCI reviews, that give bond strength data for 15 dentin-bonding agents of the so-called third and fourth generations. Seventeen selected parameters were classified into four groups: Group A includes factors related to the dentin substrate (i.e., nature of teeth); group B, composite and bonding area (i.e., composite stiffness); group C, storage conditions of the bonded samples (i.e., thermocycling); and group D, test design (i.e., crosshead speed). For each report, the experimental features, the bond strength means and standard deviations, and the failure mode were extracted and tabulated. Statistical Analysis System software was used to perform Pearson correlation analysis and analysis of variance, with bond strength as the dependent variable and experimental conditions as the independent variables. The meta-analytical review highlighted the significant influence of various parameters in the different groups: origin of dentin, types of teeth, pulpal pressure, tooth storage temperature, maximum storage time of teeth, and dentin depth in group A; type and stiffness of composite and bonding area in group B; storage of bonded samples (medium, temperature, and time) in group C, and testing mode and crosshead speed in group D. A significant positive correlation was observed between the mean bond strength and the rate of cohesive failure. It can be concluded from this study that some of these parameters should be controlled by the use of a standardized protocol. Unfortunately, the substrate-related variables are more difficult to control, even though their influence is consistent.