Repeatability and Reproducibility of a Modified Lekholm and Zarb Bone Quality Classification Based on Cone Beam Computed Tomography: An Observatsion Study

AbstractAim:We propose a modified Lekholm and Zarb classification that considers all possible combinations of cortical and cancellous bone to provide parameters that favor greater repeatability and reproducibility.Materials and Methods:This observational and analytical study consisted of a sample of 50 cone beam computed tomography (CBCT) scans. Two new types (V and VI), three subtypes to type II (II-A, II-B, and II-C), and two subtypes to type III (III-A and III-B) were added to the original bone quality classification. The new types refer to regenerated bone (type V) and bone with some pathology (type VI). The subtypes are described as type II-A: thick cortical surrounding the abundant cancellous bone with sharp trabeculae throughout the image and presence of small and visible medullary spaces; type II-B: thick cortical surrounding the abundant cancellous bone with predominance of diffuse trabeculae in the basal bone and predominant presence of wide and visible medullary spaces; type II-C: thick cortical surrounding the abundant cancellous bone with predominance of very thick and sharp trabeculae in the basal third as well as presence of small and visible medullary spaces; type III-A: thin cortical surrounding the abundant cancellous bone with sharp trabeculae throughout the image and presence of small and visible medullary spaces; type III-B: thin cortical surrounding the abundant cancellous bone with predominance of diffuse trabeculae and presence of diffuse medullary spaces. Five dental specialists were trained in the use of the modified classification and were provided with CBCT-sectioned images of edentulous jaws for classification. Each specialist classified the images twice at a 7-day interval. The strength of intra-examiner and inter-examiner agreement was measured with Cohen’s and Fleiss’ kappa index, respectively. In addition, the agreement between both classifications was analyzed. All data were analyzed at a 95% confidence level, considering a P-value