Page 75 - Peri-implant health: the effect of implant design and surgical procedure on bone and soft tissue stability
P. 75

                                LONG-TERM EFFECT OF SURFACE ROUGHNESS AND PATIENTS' FACTORS ON CRESTAL BONE LOSS AT DENTAL IMPLANTS. A SYSTEMATIC REVIEW AND META-ANALYSIS
the heterogeneity estimates of the random effects model were included and the results of meta-regression models predicting mean bone loss through mediating factors, such as smoking and roughness were presented. The aim of meta-analysis is the recombination of results of several studies through increased power and to detect influences that otherwise would appear to be statistically insignificant. Extreme heterogeneity of the studies does not allow for the straightforward application of meta-analytic statistics and testing strategies. Therefore, a subgroup analysis was conducted for roughness, smoking, and periodontal history based on studies that allowed for “paired” comparisons. Null-hypotheses were the equality of bone loss between minimally rough and moderately rough implant surfaces, smokers and non-smokers, and healthy patients and patients with a periodontal history, respectively. Throughout the meta-analytic analysis the study results are weighted by the inverse of the variance of the effect sizes. Heterogeneity was considered to be high if I2 exceeded 70%, between 50% and 70% the heterogeneity was considered to be medium, and below 50% small. When the heterogeneity was revealed to be statistically significant (p < .05) the random effects model was interpreted. If not significant, the fixed model was evaluated.
Descriptive statistics and the recombination of results were performed using MatLab R2015b version (8.6.0.267246) (The MathWorks, Inc., Natick, MA, USA). The meta-regression and fixed and random effects modeling were performed using the statistical package R version 3.1.0 (2014-04-10) (The R foundation for Statistical Computing), platform: x86_64-w64-mingw32/x64 (64-bit) with “metaphor” package (version 1.9-7) for meta-regressions and the package “meta” version 4.3-2 for random and fixed effect modeling.
■ RESULTS
Selection of Papers
The literature search yielded 6,445 studies starting in 1972 up to 2015. Of the last 2,566 publications between 2011 and August 14, 2015, in total 156 were deemed appropriate and selected by the two examiners taking the initial selection criteria into account. A further selection after reading of the paper disqualified another 69 papers for several reasons, among others the lack of standard deviation on bone loss. This was an essential factor to allow statistical estimation of the proportion of implants with an arbitrarily selected threshold for bone loss (Table 2).
73
 4



























































































   73   74   75   76   77