Analysis of reliability, accuracy, sensitivity and predictive value of a subjective method to classify facial pattern in adults

Abstract

ABSTRACT Introduction: Craniofacial pattern diagnosis is vital in Orthodontics, as it influences decision-making regarding treatment options and prognosis. Capelozza Filho proposed a subjective method for facial classification comprising five patterns: I, II, III, Long Face and Short Face. Objective: To investigate the accuracy of a subjective classification method of facial patterns applied to adults. Methods: A sample consisting of 52 adults was used for this study. Frontal and lateral view photographs were taken with subjects at rest position, including frontal smile. Lateral cephalometric radiographs were organized in a PowerPoint® presentation and submitted to 20 raters. Method performance was assessed by examining reproducibility with Kappa test and calculating accuracy, sensitivity and positive predictive values, for which 70% was set as critical value. The gold standard of the classification was personally set by the author of the method. Results: Reproducibility was considered moderate (Kappa = 0.501); while accuracy, sensitivity and positive predictive values yielded similar results, but below 70%. Conclusions: The subjective method of facial classification employed in the present study still needs to have its morphological criteria improved in order to be used to discriminate the five facial patterns.