The aim of this investigation was to evaluate the contribution of cephalometry to a statistical model integrating clinical, physical, and oximetric variables, to reduce demands for polysomnographies. Two hundred and twenty-five consecutive patients that had been referred to the sleep clinic for suspected obstructive sleep apnea (OSA) were studied. The clinical assessment of all patients consisted of a sleep related questionnaire, the Epworth sleepiness scale, and a physical examination. In addition, they all underwent spirometry, cephalometry, and a full polysomnography. The clinical variables related with OSA were questions concerning witnessing of apneas by bed partners, intensity of snoring, a history of hypertension, and nocturia. A significant relation was also found with score on the Epworth scale, sex, age, body mass index, neck and waist circumferences, total number and frequency of oxygen desaturations, and the lowest oxygen saturation value. Significant cephalometric measurements were: the linear distance from gonion to gnathion, from the hyoid bone to the mandibular plane, and from the posterior nasal spine to the tip of the soft palate, and the thickness of the uvula as well. A statistical model was built to estimate a patient's probability of having OSA based on clinical variables, physical examination, pulse oximetry, and cephalometry. The validation of this model demonstrated a remarkable ability in reducing the number of polysomnographic studies. We conclude that cephalometry combined with clinical variables, physical examination, and nocturnal oximetry is useful in the diagnosis of OSA and enables the sparing of a considerable number of polysomnographies.
