We appreciate the interest and comments of Dr Sabour regarding our recent article “Prediction of 3-dimensional pharyngeal airway changes after orthognathic surgery: A preliminary study” (Brunetto DP, Velasco L, Koerich L, Araújo MT. Am J Orthod Dentofacial Orthop 2014;146:299-309).
Dr Sabour expressed concern with our use of the Pearson correlation for predicting the upper airway changes after surgical jaw displacement. We must agree that this correlation is not useful when creating prediction models. However, we did not use this statistical method for that purpose. In fact, linear regression analysis was applied. Coefficient of determination (R 2 ), given by linear regression modeling, defines how much of the variance of the predicted variable can be explained by the independent variable variation. In addition, in our study, the causality is clear: the surgical displacement caused the variations of the upper airway dimensions—and not the other way around. For those reasons, this method can be used to create prediction models, which can be subsequently tested for efficiency, accuracy, and stability.
We understand that elaborating 2 different cohort data sets for further cross-validation might be considered the gold standard in terms of prediction, yet this method demands a much larger sample, which can derail the study. Furthermore, because this was a preliminary study, splitting the sample was not a feasible alternative because the “n” would not achieve the sample calculation’s requirement. We are constantly working to increase this casuistry, and for the future Dr Sabour’s consideration seems to be the best strategy to be followed.
Eventually, our results, based on a sample calculation with power of 0.8, supported the conclusion that maxillary surgical displacement in maxillomandibular advancement was the only reliable predictor of the upper airway’s minimal cross-sectional area variation. In fact, we believe that the regression analysis performed can be useful when one is planning whether the maxilla will be advanced and how much displacement would be ideal. However, the professional should consider our data as supplementary information and use them wisely, together with esthetic and functional aspects, to define the best clinical management for each patient.