Читать книгу Vertical 2: The Next Level of Hard and Soft Tissue Augmentation - Istvan Urban - Страница 24

Logistic regression using generalized estimation equations (GEE) were conducted to assess vertical bone gain differences and the probability of complete relative bone gain according to positional variables. Models were adjusted for defect size, healing time, age, and smoking; beta coefficients, odds ratio, and 95% confidence intervals (CIs) using the Wald chi-square test were calculated. The significance level was set at 5% (α = 0.05).

A pos-thoc power analysis determined that a sample size of 65 independent sites providing 85.1% power with 95% confidence for detecting mean gains of 4.5 and 6.0 mm was significantly different using linear regression. However, as not all sites were independent, a power correction was necessary. Each patient provided a mean of 1.14 sites, and a high within-subject correlation (CCI = 0.9) was assumed, leading to a correcting coefficient of D = 1.13. Therefore, 65 dependent sites provided the same power as 58 independent ones (80.1%).

Fig 2-1 Distribution of defects (%) by site.