RESUMO
This study aimed at assessing the prevalence of alveolar antral artery (AAA) detection by CBCT, its related variables, and at describing explanatory models useful in surgical planning, by retrospective evaluation of CBCT explorations. The modelling of the probability for detecting AAA was undertaken using logistic generalized additive models (GAM). The capacity for discriminating detection/no detection was assessed by receiver operating characteristic curves. A total of 466 sinuses were studied. Univariate models showed detection probability was linked to sinus width and thickness of the lateral bony wall, together with the shape and height of the osseous crest. AAA detection probability increased steadily until the thickness of the bony wall reached 6 mm. Multivariate models resulted good discriminators for AAA detection, particularly for females, showing an area under the curve (AUC) of 0.85. Models considering patients altogether, and those including only males offered slightly lower values (AUC = 0.79). The probability of AAA detection by CBCT was influenced by gender (higher in males and for narrow sinuses) and increases with the thickness of the sinus lateral bony wall and the height of the residual alveolar ridge. Besides, and particularly for women, the thickness of the ridge at the basal level seems to improve the explanatory model for AAA detection.
