Right statistical methods should be driven to get the right answers from research studies

摘要

As argued by Giildner et al. in "Analysis of pneumatization and neurovascular structures of the sphenoid sinus using cone-beam tomography (CBT)" (1), this is the first study that has found "a correlation between the incidence of free courses of the internal carotid artery or optical nerve and the degree of sphenoid pneumatization". It is also emphasized in the discussion section that "a strong correlation between the different courses of internal carotid artery and optical nerve had been found" (1). Especially in case of challenging endoscopic vision due to massive poly-posis, bleeding, extensive soft tissue, et cetera, during the sphenoid sinus surgery, the relationship of the sphenoid sinus with the optic and the carotid canal is very important in terms of complications. This study therefore warns the surgeons about the possible complications.There are a few studies that have evaluated sphenoid sinus pneumatization and internal carotid artery or optic canal course together, however, none of them have presented a statistical correlation between the free courses of the artery or optical nerve and the degree of sphenoid pneumatization (2, 3). The present study is one of them presenting no correlation between the degree of sphenoid sinus aeration and the free course of the carotid artery and the optic canal; but reporting only a significant correlation between the free course of the carotid canal and'pneumatization of the anterior clinoids (R = 0.7899, P = 0.001) (2).It has also been noted in cadaver studies that the more pneumatized the sphenoid sinus, the more frequently the structures bulged out into the sphenoid sinus. These structures, which are existent before the sinus development, may protrude into the sinus wall in different degrees as the cavity develops or they may course freely in the sinus (2, 3). The prevalence of bulging of the surrounding structures differed according to sphenoid sinus pneumatization (3). In such studies, no correlation has been referred.We think that the main problem of the article written by Giildner et al. is an incorrect choice of statistical method. They analyzed differences between the sexes with a t-test for independent pairs. Further analyses were done by t-test for matched pairs and bivariate correlations (1).They classified the sphenoid sinus pneumatization as conchal, presellar, sellar, and postsellar types. The postsellar type was furthermore divided into types a and b (1). Thus they have five groups of the sphenoid pneumatization and these data are nominal inputs.