Prediction of obstructive sleep apnea using a morphologic score: A SAGIC Study

Introduction: Current scores for predictive screening of obstructive sleep apnea (OSA) include subjective items, potentially reducing their diagnostic accuracy. Based on a previous score (Defiandre 2015) using solely anthropometric measures and facial structures, we created and validated a novel OSA prediction score. Methods: Data from 150 German SAGIC (Sleep Apnea Global Interdisciplinary Consortium) participants were used to develop the score. Subsequently, an independent cohort of 50 German SAGIC participants was used for validation. The ability to predict severe OSA (AHI≥30), as well as to exclude OSA (AHI≤5) was evaluated. Results: The following 5 variables were used in the score: Body mass index, neck circumference, gender, a modified Friedman tongue-position scale, and waist circumference. A score result of >7 was able to predict severe OSA with a sensitivity of 82%, a specificity of 82%, and a ROC-AUC (receiver-operating-characteristic area under curve) of 0.899. A low score result Conclusion: The predictive diagnostic score using only anthropomorphic was successful in diagnosing severe OSA, being the most clinically relevant disease entity. Moreover, it was able to exclude OSA with high specificity. The results suggest that the score needs to be adjusted for gender differences.